Objective
To evaluate the impact of a school-based health behavior intervention known as Planet Health on obesity among boys and girls in grades 6 to 8.
Design
Randomized, controlled field trial with 5 intervention and 5 control schools. Outcomes were assessed using preintervention (fall 1995) and follow-up measures (spring 1997), including prevalence, incidence, and remission of obesity.
Participants
A group of 1295 ethnically diverse grade 6 and 7 students from public schools in 4 Massachusetts communities.
Intervention
Students participated in a school-based interdisciplinary intervention over 2 school years. Planet Health sessions were included within existing curricula using classroom teachers in 4 major subjects and physical education. Sessions focused on decreasing television viewing, decreasing consumption of high-fat foods, increasing fruit and vegetable intake, and increasing moderate and vigorous physical activity.
Main Outcome Measures
Obesity was defined as a composite indicator based on both a body mass index and a triceps skinfold value greater than or equal to age- and sex-specific 85th percentiles. Because schools were randomized, rather than students, the generalized estimating equation method was used to adjust for individual-level covariates under cluster randomization.
Results
The prevalence of obesity among girls in intervention schools was reduced compared with controls, controlling for baseline obesity (odds ratio, 0.47; 95% confidence interval, 0.24-0.93; P = .03), with no differences found among boys. There was greater remission of obesity among intervention girls vs control girls (odds ratio, 2.16; 95% confidence interval, 1.07-4.35; P = .04). The intervention reduced television hours among both girls and boys, and increased fruit and vegetable consumption and resulted in a smaller increment in total energy intake among girls. Reductions in television viewing predicted obesity change and mediated the intervention effect. Among girls, each hour of reduction in television viewing predicted reduced obesity prevalence (odds ratio, 0.85; 95% confidence interval, 0.75-0.97; P = .02).
Conclusion
Planet Health decreased obesity among female students, indicating a promising school-based approach to reducing obesity among youth.
THE PREVALENCE of obesity among children and youth in the United States has increased rapidly over the past 30 years. Obesity is the most common nutritional disorder among these age groups1-4 and a major cause of excess morbidity and mortality.5 This rapid increase in prevalence indicates that the Healthy People 2000 objective for adolescent obesity will not be met.6 Although the origin of obesity is complex and relates to both genetic and environmental determinants, obesity ultimately results from an imbalance of energy intake via diet relative to energy expenditure.5 Both energy intake and energy expended via physical activity are substantially discretionary and hence the foci of intervention.
It is not clear how changes in dietary intake or changes in activity levels have contributed to recent increases in obesity.7 Television viewing has been cited as one cause of increasing prevalence, based on both longitudinal8,9 and cross-sectional studies.10,11 The impact of television viewing on obesity is most likely due to both the displacement of more vigorous activities by television and effects on diet.8,12 Foods are the most heavily advertised products on children's television,13 and television viewing time is associated with between-meal snacking.14 Experimental and observational studies indicate the effects of exercise on body fat reduction.15 Some studies indicate a potential effect of dietary fat on the prevalence and treatment of obesity.5,11 The only interventions that have shown long-term effectiveness in reducing obesity have been intensive clinical programs for obese children.16 These programs require parental participation and focus on modifications in both diet and physical activity.
The rapidly increasing prevalence of obesity in the United States, however, may require a more population-based effort, in contrast to interventions focused only on high-risk (eg, already obese) individuals.17 Interventions focused on reducing obesity prevalence among young children and adolescents may be particularly important because the risk of adult obesity is increased markedly if one is obese during adolescence.18 Schools are a potentially important channel of intervention because they offer access to large populations of students and provide the opportunity to institutionalize programs in communities; however, limited school-based research has focused on reducing obesity among adolescents. School-based interventions targeting obesity have generally treated obese students, with 6 studies indicating some effectiveness.19 School-wide interventions among adolescents, in contrast, have not focused on reducing obesity but rather on reducing cardiovascular risk and have demonstrated little impact on obesity.19,20
Planet Health is a school-based intervention designed to reduce obesity in middle-school youth (grades 6-8) by altering key physical activity and dietary risk factors, including reductions in television viewing. The intervention was included within existing classes in schools to limit cost and improve replicability, beginning when students were in grades 6 and 7 and continuing over 2 school years. In our article, we describe the impact of Planet Health on obesity prevalence among boys and girls.
Planet Health interventions occurred in 5 schools located in 4 communities in the Boston, Mass, metropolitan area; the 5 control schools were located in the same communities. Recruitment of school systems to participate was based on their willingness to implement the classroom and physical education (PE) interdisciplinary curriculum, a multiethnic student population, and cooperation with random assignment of schools to the intervention or control condition.
Informed consent procedures were followed for all students. Five schools required an active consent procedure for the survey and physical measurements; parents (or guardians) needed to return a form regardless of whether they wanted their child to participate. The remaining schools used a passive consent procedure: a letter was sent to all parents describing the project, with the option to sign and return the form if they did not want their child to participate. The 5 schools with active consent included 56% of potential participants in intervention schools and 36% of potential participants in control schools. Forms were translated into 7 languages. The study was approved by the Committee on Human Subjects at the Harvard School of Public Health.
In 1995, 10 schools from 4 communities were randomly assigned (using a random number table) to either intervention (n = 5) or control (n = 5) status. Prior to randomization, we matched schools by town (pairs of schools within 4 towns were matched) or school size and ethnic composition (1 pair matched from 2 adjacent towns) to balance factors that could affect study outcomes, such as school food services, health-related curricula, or PE. The median household income of zip code areas where the schools were located averaged $36 020 among intervention schools and $34 200 among control schools, according to 1990 Census data. This median is lower than that for all households in Massachusetts in the 1990 Census ($41 000) but similar to the US figure ($33 952).21
Student intervention status was assigned based on school enrollment in fall 1995. All students in intervention schools in grades 6 and 7 in 1995 were exposed to intervention components. Youths who continued at an intervention school in the following year received 2 school years of intervention. Control schools received their usual health curricula and PE classes and none of the Planet Health program.
The primary study contrast was between students in the intervention and control schools with respect to changes from baseline (fall 1995) to follow-up (spring 1997). The primary end point at the individual level was obesity; secondary end points included self-reports of television viewing, moderate and vigorous physical activity, percent of total dietary intake from fat, servings of fruits and vegetables, and total energy intake.
Anthropometry data and student surveys were collected at the beginning of grades 6 and 7 in fall 1995, and follow-up measurements on this cohort were collected in spring 1997 (grades 7 and 8). Because obesity22 and dietary and activity patterns vary seasonally, data were collected at the same time for both intervention and control participants. Therefore, the difference in season of measurement for baseline and outcome variables should not bias estimates of intervention effect.
The planet health intervention
Planet Health was designed to reduce obesity by increasing energy expenditure while promoting key dietary behaviors consistent with dietary guidelines.23 The intervention focused on 4 behavioral changes: reducing television viewing to less than 2 hours per day24; increasing moderate and vigorous physical activity; decreasing consumption of high-fat foods; and increasing consumption of fruits and vegetables to 5 a day or more.23
We used concepts from behavioral-choice and social-cognitive theories of individual change, with a distinctive focus on reducing television viewing. Clinical research in behavioral-choice theory indicates that reducing sedentary time coincident with a dietary intervention can decrease obesity among youths who are obese.25 Other studies indicate that choice of vigorous activity among children can be enhanced by decreasing access to sedentary activities and improving access to vigorous activity,26 with similar results shown for eating behavior.27 By allowing children choice over alternative activities when television viewing time is reduced, their perceived sense of control over physical activity alternatives is increased, and this can reinforce physical activity.28 Participants in Planet Health were encouraged to "make space" for more activity in their lives by reducing television time. Both the activity and dietary components of the intervention were designed to emphasize "lifestyle" changes in behavior. Although increased fruit and vegetable consumption has not been related to obesity prevalence, a reduction in high-fat foods coupled with the motivation to try less energy-dense fruits and vegetables could reduce obesity risk.
Social-cognitive theory29 points to the importance of social and environmental factors that influence both psychosocial and behavioral risk factors30,31 for obesity. Prior school-based interventions grounded in social-cognitive theory have produced modest reductions in dietary fat intake and increases in physical activity32-35 but little evidence for change in obesity19,20 and no evidence for reductions in television viewing. Planet Health was designed to provide students with cognitive and behavioral skills to enable change in target behaviors, practice using skills to strengthen perceived competence in employing new behaviors effectively, and support for behaviors by multiple classroom and PE teachers.
A distinctive aspect of Planet Health is the interdisciplinary curriculum approach,36 with intervention material infused into major subject areas and PE, using grade- and subject-appropriate skills and competencies. This approach is designed to enhance efficiency by using classroom teachers with minimal health education training to implement the materials, and to enhance effectiveness by involving multiple classes, which often use different approaches to learning.37 Teacher training sessions were developed to enhance implementation.38
Planet Health focused on improving the activity and dietary behaviors of all students, without singling out youths who were obese for attention. This population-based approach focuses on changing the distribution of obesity in the population17 by both reducing obesity among those already obese (remission) and preventing new cases. The population approach also limits the risk of stigmatizing obese students.39
Each intervention school received the Planet Health program of teacher training workshops, classroom lessons, PE materials, wellness sessions, and fitness funds. All teachers received training from project staff either in workshops (all PE teachers, two thirds of classroom teachers) or with the field coordinator. Based on teacher interests, an average of 3 teacher/staff wellness sessions were offered per school, provided at low cost by outside organizations.
Planet Health materials incorporate standards outlined in the Massachusetts Curriculum Frameworks,40 so that skills and competencies that are required learning in middle school are used as vehicles for conveying Planet Health messages. In classroom lessons, each theme was addressed in 1 lesson per subject (language arts, math, science, and social studies), for a total of 16 core lessons each in year 1 and year 2 (32 total). An additional lesson developed a 2-week campaign to reduce television viewing in households ("Power Down"). Lessons consisted of teacher resources; behavioral and learning objectives; procedure, extension, or homework activities; and student resources and handouts. Each classroom lesson had a behavioral objective that fit with 1 of the 4 behavioral targets of the intervention.
Units were developed with extensive teacher input via lesson evaluations and focus groups using a variety of innovative, student-centered teaching methods to engage students, including demonstrations, debates, case studies, group projects, games, and student presentations. Classroom lessons were designed for one or two 45-minute periods, depending on the level of detail chosen by teachers.
Physical education materials focused on activity and inactivity themes and included student self-assessments of activity and inactivity levels and goal setting and evaluations for reducing inactivity, replacing inactive time with moderate and vigorous physical activities of their choosing. Lessons were organized into thirty 5-minute microunits that were designed to be repeated with extensions in school-year 2; the first 5 microunits focused on Fit-Check self-assessments and goal setting. Fitness Funds were monetary incentives of $400 to $600 provided to intervention schools in response to teacher-submitted proposals consistent with Planet Health themes.
Primary End Point—Obesity. Student height without shoes was measured to the nearest 0.1 cm using a Shorr stadiometer (Shorr Productions, Olney, Md) and weight in light clothes was measured to the nearest 0.1 kg on a portable electronic scale (Seca Model 770; Seca Scale Corp, Munich, Germany) calibrated using the Seca standard weights step-up test. Triceps skinfolds (TSF) were measured to the nearest 0.2 mm using calibrated Holtain calipers. Measurements were obtained by trained project staff and standardized according to accepted protocols41 that were modified for administration in a public school setting. Replicate measures of TSF were obtained to improve precision; if the 2 measurements differed by more than 2 mm, a third measurement was taken. Analyses used the average of the measures.
We defined obese students using a composite indicator42,43 based on both a body mass index (BMI, kg/m2) and a TSF measure greater than or equal to the 85th percentile of age- and sex-specific reference data.44 Secular increases in child and adolescent obesity in the past 2 decades have been documented with indicators based on either TSF1,2 or BMI.3,4 Research has indicated, however, the limitations of using only BMI or only TSF as an obesity indicator. Among children and youth, the BMI correlates well with laboratory measures of the total and percent body fat,45,46 but differences are noted by sex, race/ethnicity, and maturation.43,47,48 The BMI has been shown to be highly specific in identifying adolescents with a high percentage of body fat, but it has low sensitivity.48 Furthermore, BMI is not a direct measure of total or percent body fat but is influenced by frame size. Changes in BMI may reflect lean body mass rather than fatness.49 Triceps skinfolds constitute a direct, valid measure of adiposity45,50 with substantial reliability,51 although unreliability tends to increase among the most overweight individuals,52 and TSF may not reflect differences in subcutaneous fat distribution that vary by racial/ethnic group and sex.47,53
Recent guidelines on consensus indicators that addressed these methodological issues in both domestic42 and international43 settings recommend a combination of indicators. One recommended composite indicator for this purpose is based on 3 measures: BMI, TSF, and subscapular skinfolds43 relative to reference standards.44 We were unable to secure approval to obtain subscapular skinfold measurements in a school setting. The composite indicator we selected to classify obesity in our multiethnic school-based sample, BMI and TSF at or above the 85th percentile, addresses methodological concerns related to indicators based on either measure alone. The combination of BMI and TSF is designed to distinguish individuals who are overweight but not overfat from the truly obese.
Sexual maturity ratings are recommended to interpret and control for differences among individuals in early and later maturational tempo that are not reflected in reference growth curves for BMI and TSF.43,47,54 In this research setting, we obtained self-reports of menarcheal status and age at menarche among girls. Use of self-reported or clinical sexual maturity ratings was not permitted by the school systems.
Food and Activity Survey.Television, Activity, and Diet Measures. Measures of television viewing, physical activity, and dietary intake as well as other sociodemographic and behavioral variables were obtained via an optically scannable student Food and Activity Survey. Students completed the Food and Activity Survey independently in class under the supervision of teachers who participated in hour-long training sessions prior to administration.
Television and Video. Hours of television and video viewing were assessed with an 11-item Television and Video Measure. Questions asked about hours of television typically viewed during each day of the week as well as use of videocassette recorders and video and computer games. Items were appropriately weighted and summed to obtain a total hours-per-day viewing estimate. Studies have indicated high test-retest reliability of youth self-reports of television viewing.55 We selected a random sample of 53 students who underwent two 24-hour physical activity recalls administered by trained interviewers 1 month apart, as well as the Food and Activity Survey. We found a deattenuated56 correlation of television viewing assessed by the Television and Video Measure with 24-hour recall of r = 0.54, with equivalent means.
Physical Activity. Moderate and vigorous physical activity levels were assessed using a Youth Activity Questionnaire, which contains 16 items that estimate hours per day spent in moderate and vigorous activities (≥3.5 metabolic equivalents)57 over the past month. Walking was not included because of low validity found for this domain.58 The Youth Activity Questionnaire was adapted from a 14-item physical activity questionnaire demonstrated to have excellent reproducibility and validity in adults.59,60 In the repeat 24-hour validation sample, the Youth Activity Questionnaire was correlated (deattenuated56) with 24-hour recall at r = 0.80, with equivalent means.
Dietary Intake. Assessment of intake of fruits and vegetables, percentage of energy from fat and saturated fat, and total energy intake (in kJ) was made using the Youth Food Frequency Questionnaire, which has been adapted and validated for use in ethnically and socioeconomically diverse populations.58,61,62 To be consistent with National Cancer Institute nutrition education guidelines, when calculating fruit and vegetable intake, we excluded french-fried potatoes.63 We excluded observations with implausible daily energy intakes (≤2100 or ≥29 000 kJ; less than 1% of observations) in dietary analyses.
Sociodemographic Variables. Age was calculated based on birth date and date of examination, except for a few cases in which the birth date was missing and Food and Activity Survey data were used. Sex was classified at the time of examination by measurers; in the few cases in which sex data were missing, school-list data were used. Ethnic categories were based on responses of students to a question asking them to mark all that apply: "How do you describe yourself?" The response categories were white, black, Hispanic, Asian or Pacific Islander, American Indian or Alaskan Native, and other. Participants checking "black" were classified as African American.
Other Survey Variables. Self-reports of weight-loss behavior were adapted from national surveillance indicators,64 including dieting to lose weight, exercising to lose weight, vomiting or taking laxatives to lose weight, and taking diet pills to lose weight, and have been associated with adolescent overweight in a multiethnic sample.65
Implementation Measures. We used teacher reports of implementation, which have shown good validity compared with classroom observations.66
Because schools, not students, were randomized, the generalized estimating equation method67 for analysis of dichotomous outcomes was used to adjust for individual-level covariates under cluster randomization, with schools nested within experimental conditions, using software written for use with SAS data sets (SAS Inc, Cary, NC).68 This approach takes into account the intraclass correlation of responses within schools. The generalized estimating equation analysis also took into account the school matching in the design, including indicator variables for randomization pairs.69 Separate regressions were estimated for boys and girls because of different patterns of incremental growth in weight and height relative to skinfold velocities and differences in maturational tempo between boys and girls.43,47,48 We controlled for ethnic category because of different patterns of BMI and TSF among ethnic groups.44,47 In regressions to predict obesity at follow-up, we controlled for intervention status, indicators for randomization pairs, and known baseline predictors of obesity, including obesity, age, ethnic category, TSF, and BMI. We then tested to see if any of the following theoretically relevant variables measured at baseline added significantly to the regression equations (P<.05): menarcheal status, cigarette smoking, dietary total energy intake per day, percentage of total energy from total and saturated fat, Television and Video Measure score, hours of daily moderate and vigorous physical activity, weight-loss behaviors, and behavioral intentions to walk more, exercise more, or watch less television. Exercising to lose weight, the only variable that significantly contributed to the equation for girls, was added to the female regression models. None of the variables contributed to the male regression models.
Similar regressions were estimated to predict changes in behavioral measures (eg, Television and Video Measure). For analysis of these continuous outcomes, we used SUDAAN software (Center for Information Technology, National Institutes of Health, Bethesda, Md) to similarly estimate regressions under cluster randomization, and control for the pair-matched design; SUDAAN estimates use an implicit Taylor linearization method.68
Analyses were conducted using an intention-to-treat protocol, with participants analyzed in their original randomized condition irrespective of the number of Planet Health sessions attended.70
To assess the extent to which intervention effects could be attributed to significant changes in behavioral variables, we reestimated regressions predicting obesity change and controlled for both the baseline behavioral variable and change in that variable from baseline to follow-up. We examined Television and Video Measure scores, fruit and vegetable intake, and total energy intake.
We used indicator variables with mean substitution to control for missing behavioral data (<5%). Because of potential bias with this approach,71 we also reestimated regressions that excluded observations with missing data. Both approaches produced similar results. In regressions estimating change in fruit and vegetable consumption, we energy-adjusted results by controlling total energy intake via linear regression.72
By design, the anthropometric measures worked in both intervention and control schools, blind to intervention status, but because of scheduling logistics, 5 of the measurers29 were not perfectly balanced. To determine if these differences affected study results, we reestimated logistic regressions that predicted obesity change, adding indicators for measurers. The outcome results were unchanged.
Participation, follow-up, and baseline characteristics
At baseline in fall 1995, both anthropometry and Food and Activity Survey data were collected from 1560 students in grades 6 and 7 in the 10 intervention and control schools. The overall participation rate was 65% of eligible students (64.5% in control and 64.8% in intervention schools). We excluded students on school lists who transferred schools at baseline, were in special education classes, or were in the wrong grade. We classified as nonparticipants the 5% of eligible students who completed a Spanish-language version of the questionnaire. Consent rates were 58% among schools with active consent and 89% among schools with passive consent.
Baseline data stratified by sex revealed no significant differences among the 1560 intervention and control students in mean values of age, BMI, TSF, or obesity prevalence. There were some differences in ethnic composition; higher percentages of African American girls (17% vs 10%) and Hispanic boys (18% vs 12%) in control schools. No differences were seen for other ethnic categories by sex. Major reasons for nonparticipation in baseline anthropometry were lack of parental consent (80% of those not participating) and absence from school (14% of those not participating).
Outcome anthropometric data were collected 2 school years later, in spring 1997, for 83% of the baseline sample. For girls, follow-up data were obtained for 82% of control and 81% of intervention students; for boys, rates were 86% and 83%, respectively. Among girls, there were no significant differences in rates of follow-up at baseline. For boys who were obese, a lower rate of follow-up was observed in the intervention condition (87% vs 94% in the controls). Main reasons for lack of follow-up anthropometric data included school transfer (52%), school absence (27%), and child refusal (10%).
For analysis of primary outcomes, the Planet Health cohort was defined as those students with both baseline anthropometry and survey data and follow-up anthropometry (n = 1295). Table 1 presents the baseline characteristics of cohort participants in fall 1995. Intervention and control participants had similar baseline sociodemographic characteristics, anthropometric data, diet, and physical activity, although some small differences remained. The prevalence of obesity was 28% in control and 27% in intervention students. Among girls, there was a difference in the prevalence of African American students (16% control vs 10% intervention).
Eighty-seven percent of the classroom teachers (n = 86) and 100% of the PE teachers (n = 9) completed first-year training sessions. Classroom teacher lesson evaluations (N = 230) indicated that they completed, on average, 3.5 lessons for the year. Teachers were expected to do a minimum of 4 lessons, and "Power Down" could be one of these. Physical education teachers completed, on average, 8.2 microunits during the year. Implementation of microunits was influenced by when the units were started during the school year, teacher motivation, and the fact that some schools had fewer PE classes (once a week or less). Qualitative data collected from teachers indicated that schools varied in the ease with which Planet Health materials were implemented. Schools experienced with interdisciplinary curricula found it easier to implement the materials.
Obesity prevalence among female students in the control schools increased from 21.5% to 23.7% over the 2 school-year intervention periods, while in the intervention schools prevalence declined from 23.6% to 20.3% (Table 2). After controlling for baseline covariates, obesity prevalence among female students in the intervention schools was significantly reduced compared with female students in the control schools (odds ratio [OR], 0.47; 95% confidence interval [CI], 0.24-0.93; P = .03). Among boys, obesity declined among both control and intervention students. After controlling for covariates, there was no significant difference in outcome (OR, 0.85; 95% CI, 0.52-1.39; P = .48).
Adjusted analyses of obesity incidence and remission indicated significantly greater remission for female students who were obese in intervention vs control schools (OR, 2.16; 95% CI, 1.07-4.35; P = .04) and no difference for boys (OR, 1.37; 95% CI, 0.44-4.24; P = .54).
Although school-specific analyses were limited by small samples, an examination of change in adjusted obesity rates by school indicated positive evidence for intervention effect among girls in 4 of the 5 pairs of schools.
We examined evidence for impact of the intervention on the 4 behavioral targets of Planet Health as well as total energy intake. Both crude and adjusted changes in these measures are displayed in Table 3, stratified by sex. After adjusting for baseline covariates, the number of television hours per day among girls was reduced in the intervention schools compared with students in control schools (−0.58 hours; 95% CI, −0.85 to −0.31 hours; P = .001). Likewise among boys, there was evidence for a reduction in television time among intervention students compared with controls (−0.40 hours; 95% CI, −0.56 to −0.24 hours; P<.001). Among girls, there was less of an increase in estimated energy intake per day over the 2 school years among intervention participants compared with controls (−575 J; 95% CI, −1155 to 0 J/d; P = .05) and an increase in fruit and vegetable consumption (0.32 servings/d; 95% CI, 0.14-0.50 servings/d; P = .003). There was no evidence for statistically significant changes in the other measures.
Behavioral Variables as Explanations for Intervention Effect
We hypothesized that change in behavioral variables (television viewing, fruit and vegetable consumption, and energy intake) could explain the intervention effect among girls. Regression results indicated that only change in television viewing mediated the intervention effect. When included in the regression predicting change in obesity along with baseline television viewing, each hour of reduction in television viewing was independently associated with a reduction of obesity prevalence (OR, 0.85; 95% CI, 0.75-0.97; P = .02), and the intervention effect was then only marginally statistically significant (P = .08, regression not shown).
Among girls who were obese at baseline, when we controlled for baseline television viewing and change in television viewing, each hour of reduction in television viewing was independently associated with increased remission of obesity (OR, 1.92; 95% CI, 1.37-2.70; P = .002), and the estimate of intervention effect was reduced (adjusted OR, 2.4-1.6) and was not statistically significant (P = .17).
Change in Obesity by Ethnic Group
We examined evidence for reductions in obesity among different ethnic groups, with samples large enough to estimate obesity rates (a minimum cell size of 5), although for all groups other than whites the numbers were small. The largest intervention effects were seen among African American girls, with obesity prevalence significantly reduced in intervention vs control participants (OR, 0.14; 95% CI, 0.04-0.48; P = .007). The intervention effect among white girls was similar to the overall result (OR, 0.46; 95% CI, 0.19-1.12; P = .08). Among Hispanic girls, results were statistically insignificant (OR, 0.38; 95% CI, 0.03-5.3; P = .42), with too few observations to examine other ethnic groups. No differences were found for boys.
We examined measures of extreme dieting behavior at both the baseline and follow-up periods to assess whether the intervention could have produced unintended side effects. Overall, students in the intervention and control schools reported similarly low levels of extreme dieting behavior at both baseline and follow-up measurements.
Planet Health is, to our knowledge, the first randomized controlled field trial of a school-wide program specifically aimed at reducing obesity among middle-school youth. Study results indicate success in reducing obesity among girls, with no significant differences observed among boys. Reductions were found in students' television viewing time as a result of Planet Health for both boys and girls, and girls in the intervention schools experienced increases in fruit and vegetable consumption and a reduced increase in dietary energy intake over the 2 school years. Among girls, reductions in television viewing time predicted reductions in obesity, mediating the intervention effect and providing evidence for this causal pathway.
The lack of an intervention effect among boys suggests that different causal factors may operate among boys and girls, although there is little published scientific evidence to support this hypothesis. Alternatively, girls could be more attuned to issues of diet and activity and thus more responsive to the intervention.73 We found a similar prevalence of reported extreme dieting behavior among both boys and girls, as noted in prior studies,65,74 although boys are much more likely to report trying to gain weight and girls to report trying to lose weight.65 We found no evidence for increases in extreme dieting behavior among boys or girls, indicating that the intervention was not having this unintended side effect.
The behavioral data validate the potentially important role of television viewing time in changing obesity, as suggested by both prior epidemiologic research8-11 and experimental results based on behavioral-choice theory.25 These results indicate that a focus on reducing television viewing time can be a useful addition to intervention efforts.
The intervention also appeared to reduce the increase in energy intake among girls. The magnitude of the effect observed (−575 J/d; 95% CI, −1155 to 0 J/d), controlling for nonbasal energy expenditure (activity), is enough to influence obesity over the course of a year. The estimated mean intake among girls in the intervention schools (8156.4 J in 1997) is consistent with recent national estimates in this age range.75
While the analysis of obesity incidence and remission indicates statistically significant intervention effects only for remission, it is important to note that over this 21-month period only 33 incident cases occurred—13 among intervention girls—which limited the statistical power of our study to detect differences. Our finding of a significant effect on remission of obesity indicates that intervention effects may be largest among those most at risk for obesity.
This intervention study had several limitations. At baseline, the participation rate of students was 65%, due in part to lower rates of participation for students whose schools required active (written) consent; however, there were no differences in participation between intervention and control schools, and baseline data revealed similar characteristics of students in the intervention and control groups as well as similar follow-up rates in both groups.
The randomization of students was limited to randomization by schools (the clusters), increasing the risk of an unbalanced design and allowing the possibility for clustering of outcome observations within schools. The cluster randomization produced very comparable intervention and control groups, and outcome analyses took the clustering into account in estimating differences among intervention and control schools, using the generalized estimating equation and SUDAAN estimation approaches.67,68
A methodological concern was the potential for measurement error in the assessment of obesity. We used a composite indicator that should minimize errors and found no evidence for bias due to measurer variability. We were unable to adjust for maturation in boys but found no evidence for differences in variables associated with maturation in intervention students vs controls.
Another measurement concern is the limited validity of measures of dietary intake and physical activity based on self-reports. Our evaluation shows significant changes in television viewing, an increase in fruit and vegetable intake, and a smaller increment in total energy intake among girls in intervention schools compared with control schools. Are changes in these variables entirely responsible for the intervention effect? Our analyses indicate this was not the case; controlling for changes in these variables only marginally reduced the OR that indicated an intervention effect, signaling that either better measures of these variables or other variables were responsible for a substantial portion of the intervention effect.
Finally, further research is needed to evaluate the extent to which these changes persist over a longer period.
The Planet Health intervention decreased obesity prevalence among female students over 2 school years. The success of this program implemented in public schools by regular classroom and PE teachers indicates a promising approach to reducing obesity among youth.
Editor's Note: The difference in outcome for boys vs girls is more proof that the sexes are from different planets. I guess Planet Health is more like Venus than Mars.—Catherine D. DeAngelis, MD
Accepted for publication September 3, 1998.
This study was supported in part by grant HD-30780 from the National Institutes of Child Health and Human Development, Bethesda, Md (Dr Gortmaker) and Prevention Research Center Grant U48/CCU115807 from the Centers of Disease Control and Prevention, Atlanta, Ga (Dr Gortmaker).
Presented in part at the Annual Meeting of the Ambulatory Pediatric Association, New Orleans, La, May 4, 1998.
We acknowledge the crucial assistance of our intervention staff, including Wendy Santis, MS, Mary Kate Newell, and Jill Carter, MEd. Special thanks to the chair of our Advisory Committee, William Dietz, MD, PhD, and to Graham Colditz, MD, DrPH, Alison Field ScD, and Helaine Rockett, RD, for assistance with nutrient analyses.
Corresponding author: Planet Health, Harvard Center for Children's Health, Harvard School of Public Health, 677 Huntington Ave, Boston, MA 02115.
1.Gortmaker
SLDietz
WH
JrSobol
AMWehler
CA Increasing pediatric obesity in the United States.
AJDC. 1987;141535- 540
Google Scholar 2.Ross
JGPate
RRLohman
TGChristenson
GM Changes in the body composition of children.
J Phys Educ Recreation Dance. 1987;5874- 77
Google ScholarCrossref 3.Shear
CLFreedman
DSBurke
GLHarsha
DWWebber
LSBerenson
GS Secular trends of obesity in early life: the Bogalusa Heart Study.
Am J Public Health. 1988;7875- 77
Google ScholarCrossref 4.Troiano
RPFlegal
KMKuczmarski
RJCampbell
SMJohnson
CL Overweight prevalence and trends for children and adolescents: the National Health and Nutrition Examination Surveys, 1963 to 1991.
Arch Pediatr Adolesc Med. 1995;1491085- 1091
Google ScholarCrossref 6.US Dept of Health and Human Services, Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC Public Health Service1990;
7.Gortmaker
SLDietz
WHCheung
L Inactivity, diet and the fattening of America.
J Am Diet Assoc. 1990;901247- 1255
Google Scholar 8.Dietz
WHGortmaker
SL Do we fatten our children at the TV set? obesity and television viewing in children and adolescents.
Pediatrics. 1985;75807- 812
Google Scholar 9.Gortmaker
SLMust
ASobol
AMPeterson
KColditz
GADietz
WH Television viewing as a cause of increasing obesity among children in the United States, 1986-1990.
Arch Pediatr Adolesc Med. 1996;150356- 362
Google ScholarCrossref 10.Shannon
BPeacock
JBrown
MJ Body fatness, television viewing and calorie-intake of a sample of Pennsylvania sixth grade school children.
J Nutr Educ. 1991;23262- 268
Google ScholarCrossref 11.Obarzanek
ESchreiber
GBCrawford
PB
et al. Energy intake and physical activity in relation to indexes of body fat: the National Heart, Lung, and Blood Institute Growth and Health Study.
Am J Clin Nutr. 1994;6015- 22
Google Scholar 12.Dietz
WHBandini
LGMorelli
JAPeers
KFChing
PLYH Effect of sedentary activities on resting metabolic rate.
Am J Clin Nutr. 1994;59556- 559
Google Scholar 13.Taras
HLGage
M Advertised foods on children's television.
Arch Pediatr Adolesc Med. 1995;149649- 652
Google ScholarCrossref 14.Clancy-Hepburn
KHickey
AANevill
G Children's behavior responses to TV food advertisements.
J Nutr Educ. 1974;693- 96
Google ScholarCrossref 15.Gutin
BManos
TM Physical activity in the prevention of childhood obesity: physical activity in the prevention of childhood obesity.
Ann N Y Acad Sci. 1993;699115- 126
Google ScholarCrossref 16.Epstein
LHValoski
AWing
RRMcCurley
J Ten-year follow-up of behavioral, family-based treatment for obese children.
JAMA. 1990;2642519- 2523
Google ScholarCrossref 18.Whitaker
RCWright
JAPepe
MSSeidel
KDDietz
WH Predicting obesity in young adulthood from childhood and parental obesity.
N Engl J Med. 1997;337869- 873
Google ScholarCrossref 19.Resnicow
K School-based obesity prevention: population versus high-risk interventions.
Ann N Y Acad Sci. 1993;699154- 166
Google ScholarCrossref 20.Sallis
JFChen
AHCastro
CM School-based interventions for childhood obesity. Cheung
LWYRichmond
JBeds.
Child Health, Nutrition, and Physical Activity. Champaign, Ill Human Kinetics1995;179- 203
Google Scholar 21.US Bureau of the Census, Statistical Abstract of the United States: 1996. Washington, DC US Bureau of the Census1998;
22.Dietz
WHGortmaker
SL Factors within the physical environment associated with childhood obesity.
Am J Clin Nutr. 1984;39619- 624
Google Scholar 23.US Dept of Agriculture, Nutrition and Your Health: Dietary Guidelines for Americans, 1995. 4th ed. Washington, DC US Dept of Agriculture1995;25Home and Garden Bulletin 232.
24.American Academy of Pediatrics, Television and the Family. Elk Grove Village, Ill American Academy of Pediatrics1986;
25.Epstein
LHValoski
AMSmith
JA
et al. Effects of decreasing sedentary behavior and increasing activity on weight change in obese children.
Health Psychol. 1995;141- 7
Google ScholarCrossref 26.Epstein
LHSmith
JAVara
LSRodefer
JS Behavioral economic analysis of activity choice in obese children.
Health Psychol. 1991;10311- 316
Google ScholarCrossref 27.Lappalainen
REpstein
LH A behavioral economics analysis of food choice in humans.
Appetite. 1990;1481- 93
Google ScholarCrossref 28.Vara
LSEpstein
LH Laboratory assessment of choice between exercise or sedentary behaviors.
Res Q Exerc Sport. September1993;64356- 360
Google ScholarCrossref 29.Bandura
A Social Foundations of Thought and Action. Englewood Cliffs, NJ Prentice-Hall Int Inc1986;
30.Perry
CLStone
EJParcel
GS
et al. School-based cardiovascular health promotion: the Child and Adolescent Trial for Cardiovascular Health (CATCH).
J Sch Health. 1990;60406- 413
Google ScholarCrossref 31.Perry
CLParcel
GSStone
E
et al. The Child and Adolescent Trial for Cardiovascular Health (CATCH): overview of the intervention program and evaluation methods.
Cardiovasc Risk Factors. 1992;236- 44
Google Scholar 32.Killen
JDTelch
MJRobinson
TNMaccoby
NTaylor
CBFarquhar
JW Cardiovascular disease risk reduction for tenth graders: a multiple-factor school-based approach.
JAMA. 1988;2601728- 1733
Google ScholarCrossref 33.Kelder
SHPerry
CLKlepp
K Community-wide youth exercise promotion: long term outcomes of the Minnesota Heart Health Program and the Class of 1989 Study.
J Sch Health. 1993;63218- 23
Google ScholarCrossref 34.Kelder
SHPerry
CLLytle
LAKlepp
KI Community-wide youth nutrition education: long-term outcomes from the Minnesota Heart Health Program and the Class of 1989 Study.
Health Educ Res. 1994;9119- 131
Google ScholarCrossref 35.Luepker
RVPerry
CLMcKinlay
SM
et al. CATCH Collaborative Group, Outcomes of a field trial to improve children's dietary patterns and physical activity: the Child and Adolescent Trial for Cardiovascular Health (CATCH).
JAMA. 1996;275768- 776
Google ScholarCrossref 36.Clark
DCClark
SN Interdisciplinary curriculum: meeting the needs of young adolescents.
Schools in the Middle. 1994;34- 7
Google Scholar 37.Gardner
H The Unschooled Mind: How Children Think and How Schools Should Teach. New York, NY Basic Books Inc Publishers1991;
38.Ross
JGLuepker
RVNelson
GDSaavedra
PHubbard
BM Teenage health teaching modules: impact of teacher training on implementing and student outcomes.
J Sch Health. 1991;6131- 34
Google ScholarCrossref 39.DeJong
W The stigma of obesity: the consequences of naive assumptions concerning the causes of physical deviance.
J Health Soc Behav. 1980;2175- 87
Google ScholarCrossref 40.Malden
MA Curriculum Frameworks. Dept of Education, Commonwealth of Massachusetts1997;
41.Westat
Inc National Health and Nutrition Examination Survey III: Body Measurements (Anthropometry). Rockville, Md Westat Inc1988;
42.Himes
JHDietz
WH Guidelines for overweight in adolescent preventive services: recommendations form an expert committee.
Am J Clin Nutr. 1994;59307- 316
Google Scholar 43.World Health Organization, Physical Status: The Use and Interpretation of Anthropometry. Geneva, Switzerland World Health Organization1995;WHO Technical Report Series, No. 854.
44.Must
ADallal
GEDietz
WH Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht
2) and triceps skinfold thickness.
Am J Clin Nutr. 1991;53839- 846
Google Scholar 45.Roche
AFSiervogel
RMChumlea
WCWebb
P Grading body fatness from limited anthropometric data.
Am J Clin Nutr. 1981;342831- 2838
Google Scholar 46.Revicki
DAIsrael
RG Relationship between body mass indices and measures of body adiposity.
Am J Public Health. 1986;76992- 994
Google ScholarCrossref 47.Daniels
SRKhoury
PRMorrison
JA The utility of body mass index as a measure of body fatness in children and adolescents: differences by race and gender.
Pediatrics. 1997;99804- 807
Google ScholarCrossref 48.Himes
JHBouchard
C Validity of anthropometry in classifying youth as obese.
Int J Obes. 1989;113183- 193
Google Scholar 49.Gortmaker
SLDietz
WH Secular trends in body mass in the United States, 1960-1980.
Am J Epidemiol. 1990;132194- 195
Google Scholar 50.Forbes
GAmihakimi
GH Skinfold thickness and body fat in children.
Hum Biol. 1970;42401
Google Scholar 51.Freedman
DSBurke
GLHarsha
DW
et al. Relationship of changes in obesity to serum lipoprotein changes in childhood and adolescence.
JAMA. 1985;254515- 520
Google ScholarCrossref 52.Lohman
TGRoche
AFMartorell
R Anthropometric Standardization Reference Manual. Champaign, Ill Human Kinetics Books1988;
53.National Heart, Lung, and Blood Institute Growth and Health Study Research Group, Obesity and cardiovascular disease risk factors in black and white girls: the NHLBI Growth and Health Study.
Am J Public Health. 1992;821613- 1620
Google ScholarCrossref 54.Tanner
JM Fetus Into Man. Cambridge, Mass Harvard University Press1990;
55.Robinson
TNKillen
JD Ethnic and gender differences in the relationships between television viewing and obesity, physical activity and dietary fat intake.
J Sch Health Educ. 1995;2691- 98
Google Scholar 56.Rosner
BWillett
WC Interval estimates for correlation coefficients corrected for within-person variation: implications for study design and hypothesis testing.
Am J Epidemiol. 1988;127377- 386
Google Scholar 57.Ainsworth
BEHaskell
WLLeon
ASJacobs
DR
JrMontoye
HFSallis
JF Compendium of physical activities: classification of energy costs of human physical activity.
Med Sci Sports Exerc. 1993;2571- 80
Google ScholarCrossref 58.Peterson
KEField
AEFox
KM
et al. Validation of YRBSS Questions on Dietary Behaviors and Physical Activity Among Adolescents in Grades 9-12. Report prepared for Division of Adolescent and School Health, Centers for Disease Control and Prevention, Atlanta, Ga1996;
59.Wolf
AMHunter
DJColditz
GA
et al. Reproducibility and validity of a self-administered physical activity questionnaire.
Int J Epidemiol. 1994;23991- 999
Google ScholarCrossref 60.Chasen-Taber
SRimm
EBStampfer
MJ
et al. Reproducibility and validity of a self-administered activity questionnaire for male health professionals.
Epidemiology. 1996;781- 86
Google ScholarCrossref 61.Rockett
HRFWolf
AMColditz
GA Development and reproducibility of a food frequency questionnaire to assess diet of adolescents.
J Am Diet Assoc. 1995;95336- 340
Google ScholarCrossref 62.Rockett
HRBreitenbach
MAFrazier
AL
et al. Validation of a youth/adolescent food frequency questionnaire.
Prev Med. 1997;26808- 816
Google ScholarCrossref 63.Subar
AFHeimendinger
JPatterson
BHKrebs-Smith
SMPivonka
EKessler
R Fruit and vegetable intake in the United States: the baseline survey of the Five-a-Day for Better Health Program.
Am J Health Promotion. 1995;9352- 360
Google ScholarCrossref 64.Trowbridge
FCollins
B Measuring dietary behaviors among adolescents.
Public Health Rep. 1993;108(suppl 1)S37- S41
Google Scholar 65.Field
AEColditz
GAPeterson
KE Racial/ethnic and gender differences in concern with weight and in bulemic behaviors among adolescents.
Obes Res. 1997;5447- 454
Google ScholarCrossref 66.Edmundson
EWLuton
SCMcGraw
SA
et al. CATCH: classroom process evaluation in a multicenter trial.
Health Educ Q. 1994;suppl 2S27- S50
Google Scholar 67.Zeger
SLiang
K Longitudinal data analysis for discrete and continuous outcomes.
Biometrics. 1986;42121- 130
Google ScholarCrossref 68.Shah
BVBarnwell
BGBieler
GS SUDAAN: Software for the Statistical Analysis of Correlated Data. Chicago, Ill SPSS Inc1996;
69.Grossman
DCNeckerman
HJKoepsell
TD
et al. Effectiveness of a violence prevention curriculum among children in elementary school: a randomized controlled trial.
JAMA. 1997;2771605- 1611
Google ScholarCrossref 70.Pocock
SJ Clinical Trials: A Practical Approach. New York, NY John Wiley & Sons Inc1993;
71.Jones
MP Indicator and stratification methods for missing explanatory variables in multiple linear regression.
J Am Stat Assoc. 1996;91222- 230
Google ScholarCrossref 72.Willett
WC Nutritional Epidemiology. New York, NY Oxford University Press Inc1990;
73.Stone
EJBaranowski
TSallis
JFCutler
JA Review of behavioral research for cardiopulmonary health: emphasis on youth, gender and ethnicity.
J Health Educ. 1995;26(suppl)S9- S17
Google Scholar 74.Killen
JDTaylor
CBTelch
MJSaylor
KEMaron
DJRobinson
TN Self-induced vomiting and laxative and diuretic use among teenagers: precursors of the binge-purge syndrome?
JAMA. 1986;2551447- 1449
Google ScholarCrossref 75.McDowell
MABriefal
RRAlaimo
K
et al. Energy and Macronutrient Intakes of Persons Ages 2 Months and Over in the United States: Third National Health and Nutrition Examination Survey, Phase 1, 1988-91. Hyattsville, Md National Center for Health Statistics1994;Advance Data From Vital and Health Statistics, No. 255.