• Open Access

Evaluation of the Living 4 Life project: a youth-led, school-based obesity prevention study

Authors


Dr J Utter, Epidemiology and Biostatistics, School of Population Health, University of Auckland, Private Bag 92019, Auckland 1142, New Zealand. E-mail: j.utter@auckland.ac.nz

Summary

The Living 4 Life study was a youth-led, school-based intervention to reduce obesity in New Zealand. The study design was quasi-experimental, with comparisons made by two cross-sectional samples within schools. Student data were collected at baseline (n = 1634) and at the end of the 3-year intervention (n = 1612). A random-effects mixed model was used to test for changes in primary outcomes (e.g. anthropometry and obesity-related behaviours) between intervention and comparison schools. There were no significant differences in changes in anthropometry or behaviours between intervention and comparison schools. The prevalence of obesity in intervention schools was 32% at baseline and 35% at follow-up and in comparison schools was 29% and 30%, respectively. Within-school improvements in obesity-related behaviours were observed in three intervention schools and one comparison school. One intervention school observed several negative changes in student behaviours. In conclusion, there were no significant improvements to anthropometry; this may reflect the intervention's lack of intensity, insufficient duration, or that by adolescence changes in anthropometry and related behaviours are difficult to achieve. School-based obesity prevention interventions that actively involve young people in the design of interventions may result in improvements in student behaviours, but require active support from leaders within their schools.

Introduction

Approximately 40% of adolescents in New Zealand are overweight or obese (1); among Pacific young people in New Zealand, the prevalence of overweight and obesity is 66% (1) (defined by the World Health Organization Child Growth Standards (2)). While the prevalence of obesity among New Zealand children appeared to stabilize between 2002 and 2007 (3), at least one report of secondary school students in socioeconomically disadvantaged area in Auckland demonstrated a 7% average annual increase in the prevalence of obesity between 1997 and 2005 (4). Socioeconomic and ethnic differences in the prevalence of obesity have been reported internationally (5–7), but the prevalence of overweight and obesity among Pacific Island adolescents in New Zealand is higher than many other ethnic populations of other Western countries. Likewise, the prevalence of overweight and obesity of Pacific Island young people in New Zealand appears to be much higher than young people living in the Pacific Islands (8,9).

School-based trials for obesity prevention are an extension of several decades of prevention research targeting health-compromising behaviours in young people (10). The school setting is a logical place for obesity prevention as it holds a unique influence on child and adolescent nutrition (11); adolescents can spend up to one-third of their day or more at school, and, in New Zealand, almost all children eat or drink something while they are at school (12). School-based interventions allow for both interventions that target individual behaviours and those that affect the wider school and social environment (e.g. policies). To assess the effectiveness of school-level interventions requires cluster-randomized trials, but issues in randomizing whole schools, cost, consent and attrition, and fidelity to the intervention all pose major challenges to these types of studies (10). These challenges are reflected in the limited evidence of effectiveness of school-based obesity prevention studies (13–16). Previous interventions have been predominately conducted at the primary school level, been short-term (1 year in duration), have focused on behaviour change (not environmental change), or contained numerous methodological issues (17–19).

For interventions to effectively reach young people, the programmes and activities must also be appropriate for youth. Youth participation, in partnership with adults, is an increasingly recognized and successful strategy to design and implement interventions aimed at improving health outcomes that are important to young people and their communities (20–22). For example, a youth-led, anti-tobacco media campaign is one of the few effective strategies to reduce youth tobacco use (23). Furthermore, states within the USA that have adopted campaigns modelled on youth participation have experienced the greatest declines in youth smoking rates, independent of what other traditional ‘top-down’ tobacco control measures were being implemented (such as excise taxes and clean air legislation) (24). Engaging young people in the development of interventions ensures the acceptability and appropriateness of the interventions. However, few obesity interventions have actively involved young people in the design or development of the interventions. The Teens Eating for Energy and Nutrition at School (TEENS) study engaged students, parents and school staff in advisory councils to inform school policies (25,26), but to our knowledge, no previous studies have incorporated a youth participation model as a major intervention initiative for school-based obesity prevention.

The current paper describes the evaluation of an intervention to improve nutrition and increase physical activity among a population of ethnically diverse, predominately Pacific Island, secondary school students. The intervention (Living 4 Life) was carried out in Auckland, New Zealand as part of the Pacific Obesity Prevention In Communities (OPIC) Project to reduce the prevalence of obesity among adolescents in the South Pacific Region.

Methods

Study design

The Living 4 Life intervention was conducted in six schools in South Auckland, New Zealand; four schools received the intervention and two schools were comparison schools (27,28). The four intervention schools were all in the Mangere region, where 60% of the population represent Pacific Island ethnicities and 40% are aged less than 20 years. The two comparison schools were also in South Auckland and chosen based on the similar ethnic composition and socioeconomic backgrounds of the students, while being distanced far enough to reduce contamination. The study was approved by the University of Auckland Human Participants Ethics Committee. The study was registered as a clinical trial: ACTRN12610000614099.

Baseline data were collected from all students (9–13 years) in 2005 and from new students entering school (9 years) at the start of the year in 2006. The intervention ran from early 2006 through 2008. Follow-up measurements were made among all students in years 11–13 at the end of 2008. The original statistical plan was to conduct a repeat measures analyses, but given the low proportion of students with both the baseline and follow-up measurements (<50%), a serial cross-sectional analysis plan was adopted instead. The decision to survey all older students in years 11–13 at follow-up was made in coordination with the participating schools, who felt this approach would be less disruptive. Younger students were not sampled at follow-up because they would not have had adequate exposure to the school intervention. Parental consent was received for students aged less than 16 years; students aged 16 years and older consented themselves.

Intervention

The Living 4 Life intervention objectives were developed after consultation with local community leaders, where the research team discussed local evidence relating to nutrition, physical activity and overweight and the community representatives determined the priority areas for their communities (27). This consultation process gave high value to community perspectives and partnerships; the decision-making process was fair and equitable. Following this consultative process, intervention objectives aimed: to improve dietary behaviours related to sugar-sweetened beverages and breakfast consumption; to increase physical activity during and after school; to improve the quality of foods sold at school; and to decrease television use. Intervention activities were originally planned for both church and school settings. During the baseline data collection, it was determined that the churches students attended were too diverse and not central to the intervention area. Subsequently, intervention activities were prioritized in the school setting.

Intervention activities were designed and implemented by Student Health Councils (SHCs) within each school. A description of the development and achievements of the SHCs is described elsewhere (29). In short, an intervention coordinator was employed by the research team to invite students to champion the Living 4 Life programme at school and to facilitate, not direct, the SHCs to ensure they had the resources and capacity to achieve their objectives. Following principles of youth participation and youth development (30), the intervention aimed to create opportunities for meaningful participation (e.g. students designing and implementing activities), quality relationships (e.g. between students, staff and the research team), and adopted a strengths-based approach by creating opportunities for student training and development. This led to numerous activities being launched throughout the intervention, including breakfast clubs (with physical activity), lunch-time activities, after-school dance, health weeks and combined student–staff initiatives within the school environment (e.g. rebranding the school canteen). It was also intended for the intervention to include curricula to reduce television use, but high-level support from the schools was not achieved for this. Likewise, improvements to the quality of school foods could not be addressed in three of the four intervention schools, because the school-food service was contracted to an external provider and there was no mandate to foster these improvements. The student initiatives were supported by high-level school environmental components such as: the instalment of new water fountains and distribution of drink bottles; provision of sporting equipment; resources for external providers/instructors (e.g. dance instructors); and improvements to the school canteen and eating area (in one school).

Measures

Student surveys were conducted at school, during the school day. A full description of the survey questionnaire, the anthropometric measurements and the mode of delivery is described fully in an article in this supplement by Swinburn et al. (31). Briefly, all anthropometric measurements were taken by trained research staff. Body mass index z-score (BMI-z) and the body mass index (BMI) categorizations of overweight and obesity were defined by the World Health Organization Child Growth Standards (2). Impedance was measured on the Tanita BC-418 body composition analyser (Tokyo, Japan); body fat percentage was calculated using equations validated for this population (32). The quality of life measures included the previously validated Pediatric Quality of Life Inventory (PedsQoL)™ 4.0 (generic module for 13- to 18-year-olds) developed by Dr James Varni (33) and the Assessment of Quality of Life (AQoL) (34). Key nutrition and physical activity behaviours were assessed by student self-report.

Data analysis

All data were analysed using sas v9.2 (SAS Institute Inc., Cary, NC, USA). Analyses were considered statistically significant at P < 0.05. At baseline, students in years 9 and 10 were removed from the analyses to create two similar populations for comparison. In total, only 24 students were included in both the baseline and follow-up measurements. As there were only six schools in this study, all analyses were conducted to examine changes in the outcome variables between baseline and follow-up by condition and within schools.

Anthropometric outcomes and quality of life

Unadjusted means and 95% confidence intervals were generated to describe the direction of the relationships between the intervention and comparison conditions for BMI, BMI-z, weight, body fat percentage, PedsQoL and AQoL. Adjusted analyses were conducted using mixed model regression equations, controlling for age, sex and ethnicity, with schools treated as random effects.

Behavioural outcomes

With only six schools, multilevel regression models could not be conducted with binary outcomes. Therefore, we aggregated data to the school level and ran regular anova models, controlling for baseline estimates. Unadjusted proportions and 95% confidence intervals were generated to describe the direction of the relationships between conditions.

Anthropometric and behavioural outcomes within schools

Unadjusted proportions, means and 95% confidence intervals were generated to describe the direction of the relationships between the two measurements within each school. Regression models were used to determine if the outcomes at follow-up were statistically different from those at baseline, while controlling for age, sex and ethnicity.

Results

At baseline, there were 5837 students on the school rolls and 3881 students agreed to participate in the study (response rate 66%). After the 9- and 10-year-old students were removed from the data set for analyses, the final study sample at baseline included 1634 students. At follow-up, 2515 students were invited to participate and 1612 students agreed (response rate 66%). The demographic characteristics of participating students are described in Table 1.

Table 1.  Demographic characteristics of study sample at baseline and follow-up
 BaselineFollow-up
TotalInterventionComparisonSchool 1School 2School 3School 4School 5School 6TotalInterventionComparisonSchool 1School 2School 3School 4School 5School 6
n = 1634n = 953n = 681n = 363n = 166n = 287n = 137n = 185n = 496n = 1612n = 1023n = 589n = 467n = 211n = 221n = 124n = 147n = 442
%%%%%%%%%%%%%%%%%%
Gender                  
 Male48.450.445.750.145.851.254.751.943.444.843.347.445.637.043.944.446.947.5
 Female51.649.654.349.954.248.845.348.156.755.256.752.654.463.056.155.753.152.5
Age (years)                  
 1539.139.039.141.622.944.640.240.538.537.534.343.133.833.236.234.740.843.9
 1634.234.433.838.330.732.832.133.034.135.837.133.635.639.835.840.333.333.7
 1722.422.322.616.834.919.926.323.222.423.024.320.526.322.324.919.420.420.6
 184.44.34.63.311.52.81.53.25.03.74.32.74.34.73.25.75.41.8
Ethnicity                  
 Pacific63.770.055.455.570.985.075.989.242.771.675.764.562.791.985.579.890.555.9
 Maori16.011.023.38.524.28.07.39.228.511.97.120.28.64.88.13.28.824.0
 Asian/other12.614.69.730.33.65.65.11.113.011.113.76.624.42.44.58.90.78.6
 European7.44.411.65.51.21.411.70.515.85.43.58.74.31.01.88.10.011.5

Anthropometric and behavioural outcomes by condition

Reliable measurements for BMI were collected from 1632 students at baseline and 1589 students at follow-up. For body fat percentage, reliable estimates were calculated for 1602 students at baseline and 1589 students at follow-up. In unadjusted analyses, mean BMI, BMI z-score, weight and body fat percentage all increased in the intervention group between baseline and follow-up, while slight reductions were observed in the comparison group (Table 2). However, there were no statistically significant effects of the intervention on any of the anthropometric measurements in the adjusted analyses. Similarly, there were no statistically significant intervention effects on either of the quality of life measures or on any of the behavioural outcomes.

Table 2.  Anthropometric and behavioural outcomes by condition
 Anthropometric outcomesP-value
InterventionComparison
Mean(CI)Mean(CI)
BMI
 Baseline25.36(24.7, 26.0)25.14(24.3, 26.0) 
 Follow-up25.82(25.2, 26.5)25.02(24.2, 25.9)0.18
BMI-z (WHO)
 Baseline1.02(0.9, 1.2)1.00(0.8, 1.2) 
 Follow-up1.11(1.0, 1.3)0.95(0.8, 1.5)0.13
Weight (kg)
 Baseline72.74(71.1, 74.4)72.17(70.0, 74.4) 
 Follow-up74.21(72.5, 76.0)71.91(69.6, 74.2)0.21
Body fat percentage
 Baseline31.30(29.9, 32.7)30.73(28.8, 32.7) 
 Follow-up31.82(30.4, 33.2)30.18(28.2, 32.2)0.16
PedsQoL
 Baseline79.39(78.4, 80.3)80.41(79.4, 81.4) 
 Follow-up80.29(79.3, 81.2)81.08(80.0, 82.2)0.81
AQoL
 Baseline0.73(0.7, 0.7)0.75(0.7, 0.8) 
 Follow-up0.76(0.7, 0.8)0.76(0.7, 0.8)0.09
 Behavioural outcomesP-value
InterventionComparison
%(CI)%(CI)
  1. AQoL, Assessment of Quality of Life; BMI, body mass index; BMI-z, body mass index z-score; PedsQoL, Pediatric Quality of Life Inventory.

Prevalence of obesity
 Baseline31.9(29.0, 34.9)28.5(25.1, 32.0) 
 Follow-up34.7(31.7, 37.6)30.2(26.5, 34.0)0.46
Lunch-time activity (% yes)
 Baseline29.8(26.9, 32.7)34.2(30.6, 37.7) 
 Follow-up31.7(28.8, 34.5)31.2(27.5, 35.0)0.57
After-school activity (% doing any in past 5 d)
 Baseline83.5(81.1, 85.8)83.1(80.2, 85.9) 
 Follow-up81.9(79.6, 84.3)81.5(78.4, 84.6)0.65
Television use (% using on all of past 5 d)
 Baseline41.9(38.8, 45.0)40.4(36.7, 44.0) 
 Follow-up33.9(31.0, 36.8)38.5(34.6, 42.5)0.09
Soft drink consumption (% consuming on all of past 5 d)
 Baseline20.4(17.9, 23.0)16.6(13.8, 19.4) 
 Follow-up14.2(12.0, 16.3)17.5(14.4, 20.6)0.42
Breakfast consumption (% eating on 0 of past 5 d)
 Baseline24.4(21.5, 27.3)26.8(23.3, 30.4) 
 Follow-up18.4(16.0, 20.7)19.3(16.1, 22.5)0.99
School encourages activity (% reporting a lot)
 Baseline24.7(22.0, 27.5)20.3(17.3, 23.4) 
 Follow-up32.1(29.2, 34.9)26.7(23.1, 30.2)0.63
School encourages healthy eating (% reporting a lot)
 Baseline33.7(30.5, 36.9)41.3(37.4, 45.2) 
 Follow-up47.1(43.8, 50.3)46.3(42.1, 50.7)0.74
Healthiness of school canteen (% reporting its mostly healthy)
 Baseline26.6(23.7, 29.4)35.8(32.2, 39.4) 
 Follow-up25.5(22.8, 28.2)38.2(34.3, 42.1)0.07

Anthropometric and behavioural outcomes within schools

In five of the six schools, there were no differences in any of the anthropometric measurements between baseline and follow-up, while controlling for demographic characteristics (Table 3). In school 1 (intervention), there was a significant increase in mean BMI, BMI-z, weight and body fat percentage over the study period. School 3 (intervention) was the only school where all of the anthropometric measurements appeared to decrease over the study period, but none of these estimates reached statistical significance.

Table 3.  Anthropometric outcomes of study population by school
 InterventionComparison
School 1School 2School 3School 4School 5School 6
Mean(CI)Mean(CI)Mean(CI)Mean(CI)Mean(CI)Mean(CI)
  1. Bold font indicates that the baseline and follow-up estimates were significantly different at P < 0.05, while controlling for age, sex and ethnicity.

  2. AQoL, Assessment of Quality of Life; BMI, body mass index; BMI-z, body mass index z-score; PedsQoL, Pediatric Quality of Life Inventory.

BMI
 Baseline25.17(24.5, 25.8)27.58(26.8, 28.4)28.50(27.7, 29.3)27.20(26.1, 28.3)27.53(26.8, 28.3)26.11(25.6, 26.7)
 Follow-up26.69(26.1, 27.3)28.44(27.6, 29.3)27.93(27.1, 28.7)28.50(27.4, 29.6)28.27(27.2, 29.3)26.13(25.6, 26.7)
BMI-z (WHO)
 Baseline0.99(0.8, 1.1)1.60(1.4, 1.8)1.74(1.6, 1.9)1.47(1.3, 1.7)1.64(1.4, 1.8)1.22(1.1, 1.3)
 Follow-up1.27(1.1, 1.4)1.72(1.6, 1.9)1.65(1.5, 1.8)1.73(1.5, 1.9)1.66(1.5, 1.8)1.27(1.2, 1.4)
Weight
 Baseline72.92(70.7, 75.1)80.20(77.5, 93.0)82.53(80.1, 84.9)79.88(76.1, 83.6)79.76(77.4, 82.1)75.67(73.8, 77.5)
 Follow-up77.27(75.2, 79.3)81.99(79.4, 84.6)81.12(78.6, 83.7)83.57(79.5, 87.6)82.27(78.8, 85.8)76.12(74.3, 77.9)
Body fat percentage
 Baseline29.95(28.6, 31.3)34.55(32.8, 36.3)35.37(35.5, 38.3)33.56(32.4, 36.9)33.82(34.1, 37.4)32.14(31.0, 33.3)
 Follow-up32.60(31.6, 33.6)35.92(34.5, 37.3)34.09(32.7, 35.5)34.93(33.1, 36.8)33.82(32.1, 35.6)31.46(30.4, 32.5)
PedsQoL
 Baseline78.52(77.1, 79.9)79.08(77.2, 81.0)79.14(77.7, 80.6)79.18(77.2, 81.1)79.14(77.3, 81.0)80.27(79.2, 81.3)
 Follow-up80.73(79.5, 81.9)79.50(77.8, 81.2)78.24(76.3, 80.1)77.49(75.1, 79.9)81.13(79.0, 83.3)80.34(79.2, 81.5)
AQoL
 Baseline0.73(0.7, 0.8)0.74(0.7, 0.8)0.71(0.7, 0.7)0.70(0.7, 0.7)0.73(0.7, 0.8)0.76(0.7, 0.8)
 Follow-up0.77(0.7, 0.8)0.75(0.7, 0.8)0.74(0.7, 0.8)0.73(0.7, 0.7)0.74(0.7, 0.8)0.77(0.8, 0.8)

The behavioural outcomes by school are described in Table 4 and summarized in Table 5. Consistent with the anthropometric outcomes by school, school 1 (intervention) was the only school to observe a significant increase in the prevalence of obesity (from 23% to 30%), while all other schools showed no significant differences. This may reflect a regression towards the mean, as the prevalence of obesity (23%) in school 1 was markedly lower than that observed in the other schools at baseline. In two of four intervention schools, students showed improvements related to breakfast consumption, television use and soft drink consumption. In three out of four intervention schools, students reported improvements in school encouragement for healthy eating and physical activity. These improvements were generally aligned with the intervention activities occurring within those schools.

Table 4.  Behavioural outcomes of study population by school
 InterventionComparison
School 1School 2School 3School 4School 5School 6
%(CI)%(CI)%(CI)%(CI)%(CI)%(CI)
  1. Bold font indicates that the baseline and follow-up estimates were significantly different at p < 0.05, while controlling for age, sex and ethnicity.

Prevalence of obesity
 Baseline22.7(18.4, 27.0)35.5(28.3, 42.8)39.6(33.9, 45.3)35.8(27.7, 42.8)35.68(28.8, 42.6)25.9(22.0, 29.7)
 Follow-up30.3(26.0, 34.5)39.5(32.8, 46.2)34.6(28.2, 40.9)43.1(34.3, 51.9)35.9(28.0, 43.8)28.4(24.1, 32.6)
Lunch-time activity (% yes)
 Baseline27.1(22.5, 31.7)21.8(15.5, 28.1)34.3(28.8, 39.8)37.2(29.1, 45.3)54.6(47.4, 61.8)26.5(22.6, 30.4)
 Follow-up28.9(24.8, 33.0)32.2(25.9, 38.5)40.7(34.2, 47.2)25.0(17.4, 32.6)44.9(36.8, 52.9)26.7(22.6, 30.8)
After-school activity (% doing any in past 5 d)
 Baseline79.6(75.4, 83.7)85.5(80.1, 90.8)86.7(82.8, 90.7)84.7(78.6, 90.7)88.6(84.1, 93.2)81.0(77.5, 84.4)
 Follow-up78.2(74.4, 81.9)82.9(77.9, 88.0)87.3(82.9, 91.7)84.7(78.3, 91.0)84.4(78.5, 90.2)80.5(76.8, 84.2)
Television use (% using on all of past 5 d)
 Baseline46.1(41.0, 51.3)48.5(40.9, 56.1)39.2(33.5, 44.8)28.5(20.9, 36.0)40.5(33.5, 47.6)40.3(36.0, 44.6)
 Follow-up36.6(32.2, 41.0)31.3(25.0, 37.5)32.1(26.0, 38.3)31.5(23.3, 39.6)36.7(28.9, 44.5)39.1(34.6, 43.7)
Soft drink consumption (% consuming on all of past 5 d)
 Baseline19.6(15.5, 23.7)29.7(22.7, 36.7)17.1(12.8, 21.5)18.2(11.8, 24.7)16.8(11.4, 22.1)16.6(13.3, 19.9)
 Follow-up14.1(11.0, 17.3)15.2(10.3, 20.0)16.3(11.4, 21.2)8.9(3.9, 13.9)23.8(16.9, 30.7)15.4(12.0, 18.8)
Breakfast consumption (% eating on 0 of past 5 d)
 Baseline26.3(21.4, 31.2)20.6(14.4, 26.8)25.6(20.1, 31.1)22.1(14.8, 29.5)19.5(13.5, 25.5)29.7(25.4, 34.1)
 Follow-up19.7(16.1, 23.4)12.8(8.3, 17.3)16.1(11.2, 20.9)26.6(18.8, 34.4)16.3(10.3, 22.3)20.3(16.5, 24.1)
School encourages activity (% reporting a lot)
 Baseline21.0(16.8, 25.2)18.8(12.8, 24.8)30.1(24.7, 35.4)30.7(22.9, 38.4)28.6(22.1, 35.2)17.2(13.9, 20.5)
 Follow-up28.5(24.4, 32.6)40.3(33.7, 46.9)42.5(36.0, 49.1)12.9(7.0, 18.8)30.6(23.2, 38.1)25.3(21.3, 29.4)
School encourages healthy eating (% reporting a lot)
 Baseline27.1(22.2, 32.0)20.3(13.8, 26.8)45.2(39.1, 51.4)43.7(34.8, 52.6)54.4(46.9, 62.0)36.4(31.9, 40.8)
 Follow-up45.8(41.1, 50.6)48.5(41.0, 56.1)59.3(52.4, 66.2)29.4(21.2, 37.6)48.8(40.2, 57.5)45.5(40.6, 50.4)
Healthiness of school canteen (% reporting its mostly healthy)
 Baseline30.1(25.4, 34.8)12.1(7.1, 17.1)23.9(18.9, 28.8)40.1(31.9, 48.4)39.5(32.4, 46.5)34.4(30.2, 38.6)
 Follow-up27.4(23.4, 31.5)18.5(13.2, 23.7)28.1(22.1, 34.0)25.8(18.1, 33.5)34.7(27.0, 42.4)39.4(34.8, 43.9)
Table 5.  Summary of behavioural outcomes by school
 InterventionComparison
School 1School 2School 3School 4School 5School 6
  1. +, significant improvement in school.

  2. −, significantly worse change within school.

  3. NC, no significant changes within school.

% obeseNCNCNCNCNC
Lunch-time activityNCNCNCNCNC
After-school activityNCNCNCNCNCNC
Television use++NCNCNCNC
Soft drink consumptionNC+NC+NCNC
Breakfast consumption+NC+NCNC+
School encourages activity+++NC+
School encourages healthy eating+++NC+
Healthiness of canteen foodNC+NCNCNC
Quality of life+NCNCNCNCNC

It is of interest that student perceptions of the school food and activity environment in school 4 (intervention) worsened over the intervention period. This may reflect a major senior administrative change during the intervention, which resulted in less support for the intervention. In school 6 (comparison), students were more likely to eat breakfast and report school encouragement for healthy eating and activity over the study period. This school also had senior administrators who were very supportive of nutrition and activity initiatives in the school. At baseline, we observed that the school had initiated new policies for improving the quality of school foods and beverages. Consequently, positive changes to student perceptions of the school environment and breakfast consumption were observed over the study.

Discussion

The primary aim of the Living 4 Life study was to reduce the prevalence of obesity in young people through a youth-led, school-based intervention. The Living 4 Life study is the first large school-based obesity prevention study in New Zealand targeting adolescents; previous large school-based studies in New Zealand have included only younger children (35,36). Overall, the given data suggest that a youth-led intervention (supported by environmental initiatives, including school policies) may result in positive improvements to student perceptions and eating and activity behaviours, although improvements to body size were not achieved. Taken in total, however, these findings were modest. When comparing intervention schools to the comparison schools, the intervention appeared to have no significant effect on any of the anthropometric or behavioural outcomes. Within individual schools, changes in student body weight and eating and activity behaviours were variable.

The few positive findings in our study may reflect a number of situations, but overall are consistent with the modest findings of school-based obesity prevention studies reported internationally (13–16). It is likely that while we observed some positive changes within individual schools, we did not observe a significant effect of the intervention, because although our study had large numbers of students, we had only six schools. Thus, our study may not have been powered to detect a significant difference at the school level. Even within individual schools, reductions in mean BMI and the prevalence of obesity were not achieved over the 3-year intervention. It is noteworthy that, at baseline, approximately 30% of students were obese. While some treatment interventions show some promise for younger children (37), treatment interventions are more intensive and individually focused than is possible for a school-based study.

The Living 4 Life study was designed as a school-based intervention as it is well recognized that schools are important sites for youth health promotion, including obesity prevention (11,38). Furthermore, the Living 4 Life intervention actively involved young people in the design, development and delivery of the intervention activities to ensure the appropriateness and acceptability of the interventions. Despite this, adequate acknowledgement of the wider familial, socio-cultural, economic and media environments could not be achieved within the scope of this study, regardless of these factors all being critical in addressing adolescent obesity (13,38).

Over the intervention period, there was one comparison school that demonstrated significant improvement in student behaviours and perceptions of the school environment. This may highlight the potential impact that school policies and leadership can have without an additional investment in resources. The implementation of school nutrition guidelines can result in improvements in both the quality of foods provided at school and students' dietary intakes (39). The improvements achieved by the comparison school may also reflect ‘contamination’ of obesity prevention initiatives in the wider environment. In 2004, the New Zealand government launched its national strategy for obesity prevention, subsequent to which many health promotion programmes were initiated in economically deprived areas, including the areas surrounding our comparison schools.

The Living 4 Life study was the first large study to attempt to curb the high prevalence of overweight and obesity among New Zealand young people, particularly Pacific young people. The study was unique in its implementation of a youth participation model for developing the intervention activities and its long intervention period. With these strengths, come limitations to the research that warrant consideration. First, as approximately 60% of students were overweight/obese at baseline (9), this type of intervention may be too late in the life cycle. In a longitudinal study of adolescents, few new cases of overweight or obesity emerged and few overweight young people returned to a healthy weight between the ages of 11 and 17 (40). Taken together, our findings suggest that future school-based initiatives may be more effective for younger children. Second, working with the intervention schools required flexibility around the activities to meet the needs and expectations of the schools and students. Thus, the activities themselves varied between schools and we were not able to implement all of the activities intended in all of the schools. Lastly, the general approach of a youth participatory model for obesity prevention was successfully implemented in this study, but the effectiveness of the overall approach could not be adequately tested within the existing study design. Ideally, a cluster-randomized trial with more schools would result in a higher powered study, but the costs of such an intervention were prohibitive.

Conclusion

Overall, these findings suggest that the Living 4 Life study may have resulted in positive improvements to student perceptions and eating and activity behaviours within individual schools. To achieve meaningful reductions in obesity, however, future initiatives will require more intensive interventions that acknowledge and address the wider contexts and environments in which young people live. Widespread, environmental and policy changes are necessary to effectively combat obesity and we have found that young people can make meaningful contributions to these.

Conflict of Interest Statement

M. Moodie and B. A. Swinburn's institutions have received grants, and support to cover costs of travel to New Zealand and Investigator meetings, from the National Health and Medical Research Council. The authors were employed by Deakin University.

J. Utter's institution has received grants from the Health Research Council of New Zealand. Support for travel to meetings for the study was provided by the Auckland Medical Research Foundation.

J. Warbrick, S. Foroughian, G. Faeamani, R. Scragg and E. Robinson's institutions have received grants from the Health Research Council of New Zealand.

R. Scragg has a pending grant from the Health Research Council of New Zealand.

O. Dewes was employed by University of Auckland.

Acknowledgements

The authors would like to thank the many people involved in the Pacific OPIC Project including other co-investigators, other staff and postgraduate students, partner organizations, and especially the schools, students, parents and communities. The funding for the project was from the Wellcome Trust (UK), the National Health and Medical Research Council (Australia) and the Health Research Council (New Zealand) through their innovative International Collaborative Research Grant Scheme. Additional support was also provided by AusAID.

Ancillary