Pediatric Highlight | Published:

School-based obesity prevention in Chilean primary school children: methodology and evaluation of a controlled study

International Journal of Obesity volume 28, pages 483493 (2004) | Download Citation



OBJECTIVE:To assess the impact of a 6 months nutrition education and physical activity intervention on primary school children through changes in adiposity and physical fitness.

DESIGN: Longitudinal school-based controlled evaluation study.

SUBJECTS: Children from 1st to 8th grade, 2141 in intervention and 945 in control schools.

INTERVENTION: Nutrition education for children and parents, ‘healthier’ kiosks, 90 min of additional physical activity (PA) weekly, behavioral PA program and active recess.

MEASUREMENTS: Adiposity indices (BMI, BMI Z-score, triceps skinfold thickness (TSF), waist circumference and physical fitness (20 m shuttle run test and lower back flexibility).

RESULTS: Positive effect on adiposity indices (except TSF) was observed in boys (P<0.001 for BMI Z), while both physical fitness parameters increased significantly in both boys (P<0.001 for each test) and girls (P<0.0001 for each test). A differential effect in BMI Z was observed according to baseline nutritional status.

CONCLUSIONS: This intervention showed a robust effect on physical fitness in both genders and decreased adiposity only in boys.


Chile is facing a progressive rise in obesity with the corresponding consequences in the epidemiological and nutrition profile of the population. A sustained increase in risk factors for nutrition-related chronic diseases (NR-CDs) has occurred, particularly obesity, which although has affected all age groups, in preschool and school age children, the rise in prevalence has been the greatest.1 The most probable causes include an increased urbanization and rising incomes that have led to a drastic change in fat intake, especially saturated and trans fats as well as to a progressive sedentary behavior. As stated by Popkin,2 large shifts in diet and physical activity patterns have occurred in low- and moderate-income developing countries in the last 20 y. The ‘western diet’, which is high in saturated fat, sugar and low in fiber, has rapidly replaced traditional diets. In addition, the present urban environment favors motorized vehicles, leaving little space for parks and recreational areas to facilitate an active life in the city. Also, the large income disparity and rise in number of cars have compromised personal safety in the street, so children can no longer play in public places.3

The Ministry of Health of Chile, aware of the consequences of present trends in NR-CDs (obesity, diabetes, hypertension, coronary heart disease) and related burden of disease, established in 1997 a Health Promotion Program (Vida Chile) and defined goals for physical activity and obesity throughout the life course. Specific health promotion activities are centered on five conditioning factors: healthy diet, physical activity, tobacco control, psychosocial and healthy environment. An action plan has been defined focusing on schools, the workplace and community life. The goal for childhood obesity is to reduce obesity (weight for height >2 s.d. WHO/NCHS 1977) among schoolchildren in first grade from 16% (year 2000 baseline) to 12% by year 2010, reduce sedentary behavior in those 15 y of age and older from 91 to 84% and reduce tobacco use among eighth graders from 27 to 20%.4 Vida Chile has developed an initiative called ‘Healthy Schools’, which consists in implementing actions addressing simultaneously three of the five risk factors for NR-CDs.

The implementation of school-based programs plays an important role in promoting lifelong physical activity and healthy eating in children. The emphasis of these programs should be on developing knowledge, attitudes and behavioral skills needed to establish and maintain healthy eating and an active lifestyle.

The ‘Healthy School’ initiatives have minimal evaluation components and are focused exclusively on process indicators. Since evaluations of efficacy under controlled conditions are absolutely necessary in order to develop effective programmatic actions, we developed and implemented a controlled school-based obesity prevention intervention. It included actions addressing four risk factors (diet/nutrition, physical activity, tobacco control and healthy environment). It is important to note that there are virtually no controlled intervention studies on obesity prevention in developing countries.5

We hypothesized that children exposed to a diet/nutrition and physical activity intervention during 6 months would show a significant difference in physical fitness, body mass index (BMI) and other adiposity indices compared with controls. In this paper, we describe the dietary and physical activity interventions and provide results of anthropometric and physical fitness outcomes.



Five schools were selected from three different cities: for Santiago (large city, around 5 million people) and Curico (medium-size city, population 245 000), an experimental and a comparison school were chosen; for Casablanca (small city, population 22 000), only one school met the selection criteria and was assigned to the intervention. The intervention school drawn from Santiago included 821 children (school A), from Curico 350 (school B) and Casablanca 1204 (school C). Control schools from Santiago and Curicó included 707 and 495 children, respectively. Thus, the number of children who were intervened amounted to 2375, while those in the comparison group 1202. Cities were selected by the research team to represent urban communities of different size. Schools were eligible if they met the following criteria: primary level public schools (1st–8th grades only), children with full-day school attendance (8:30–16:30), low socioeconomic status (approximately 35% of children receiving School Lunch Program) and no previous participation in health promotion programs. Assignment to intervention or control school was made by county educational authorities, considering their perception of overweight prevalence and the willingness of the school director to accept a research study. The intervention schools were thus potentially biased for increased prevalence of obesity; this was an inherent limitation, but was part of the preconditions defined by the educational authorities.

Intervention program

A trained nutritionist and physical education (PE) teacher (one of each per intervention school) were responsible for implementing the diet/nutrition education and physical activity program and evaluating the process. They were responsible for collecting anthropometric and dietary data, providing training to school teachers, supporting them during the 6- month intervention period, and conducting some of the physical activity workshops for children. All teachers in the intervention schools received on-site training for 2 days to provide them with general information on the nature and significance of the intervention. Those directly involved received additional training from the nutritionist and physical education teacher to support their role in educating the children. These activities took in the case of the nutritionist approximately 1 h per day and 2 h per week by the research team PE teacher for about 4 months.

Diet and Nutrition Intervention

This included:

  1. Educational program for children from fourth to eighth grades: We implemented a shorter version of a specially designed child-friendly classroom nutrition education program, which was recently developed by INTA and FAO6 with the objective of introducing food and nutrition contents for the curricula of Latin American primary schools. The program was initiated with 4th grade students since the dietary assessment questionnaires to be completed by the children required them to be able to read and write. Children were expected to receive approximately 8–11 h from 4th to 6th grade, and 5–6 h for 7th and 8th grades over the 6 months. The research nutritionists checked if contents were being given and activities completed by the children by occasional spot-checks.

  2. Kiosks: Schools have kiosks that sell mostly high fat, sugar and salty snacks, and in much smaller quantities, healthy foods. They are privately owned and pay a monthly rent to the school. We held two meetings with owners providing information on healthy diets and potential healthy snacks, in an attempt to influence what was being offered to children. The goals of the intervention were also explained giving ideas on what healthier food items could be sold and still remain profitable.

  3. Parental involvement: Two meetings were held with parents of children from 4th to 8th grades and were mainly directed at healthy eating, obesity prevention and to reinforce national food-based dietary guidelines.

  4. In addition, some teachers developed special activities in support of the educational program. An activity that was successfully adopted in all schools over the second half of the intervention was a contest called ‘Healthy Snack’, consisting in giving once a week a sticker to 15 children picked randomly eating a healthy snack during recess. At the end of the school year, those children who collected the most stickers received a physical activity-related item as a prize.

Physical activity intervention

This included three aspects:

  1. Canadian active living challenge (CALC): CALC is a practical behavioral resource designed to instill a healthy and active lifestyle for children aged 6–18 y.7 The objectives of this ‘hands-on’ tool is to build knowledge about the benefits and importance of health and active living as well as encourage children to increasingly incorporate activity into every-day life. CALC was translated into Spanish and adapted in terms of types of activities to be undertaken. This program was selected based on the recommendation of technical cooperation given by the University of Toronto's Center for Health Promotion to the Chilean Health Promotion program. This program is intended to be applied daily by the classroom teacher, but in this case, it was not possible to train the teachers, so the research PE teacher was responsible for its application and could only do it once a week on the children from 1st to 8th grade.

  2. Provision of an extra 90 min per week of physical activity to children from 3rd to 8th grades during 6 months: These were mainly oriented toward a certain sport (soccer, basketball and volleyball) and were conducted by the school PE teacher/classroom teacher or research team PE teacher.

  3. Active recess: During one daily recess (15 min per day), music was played at recess time, so children were encouraged to dance, play ping-pong, basketball or volleyball as recreation, using the equipment provided by the study. This activity was implemented for approximately 3 months, during the second half of the intervention period.

  4. Extra program: During the implementation of the PE program, the research team promoted activities beyond those planned originally. These were based on the individual interest of the PE teacher and varied according to the schools' facilities.

The PE equipment for the participating schools was inadequate to support the needs of the study: thus, as part of this intervention, we provided basic sports equipment, such as soccer balls, basket balls, volley balls, hula hoops, basketball boards and ping-pong tables. They were extensively used during recess time.


Primary outcome measures

These included anthropometric and physical fitness parameters. They were collected at the beginning of the school year (March/April 2002) in both intervention and control schools and repeated at the end of the school year (November).

  • Anthropometry: Weight, height, triceps skinfold thickness (TSF) and waist circumference (WC) were measured on all children by the three study nutritionists. They were trained thoroughly; intra- and interobserver reliability were determined for TSF and WC by intraclass correlation coefficient (r-value) and one-way ANOVA (Bartlett's test for equal variances), respectively. Intraclass r-values for TSF and WC were 0.92 and 0.94, while P-values from the ANOVA were 0.9995 and 0.982, respectively. Weights were taken without shoes or belts and with light clothing, and recorded to the nearest 0.1 kg with a portable digital scale (Seca model 840). Heights were measured with a standing stadiometer (Seca model 720) and recorded with a precision of 1 mm. The BMI was calculated as Wt/Ht (m).2 Children were classified as underweight (BMI<percentile or P 10), normal (BMI P 10–85), overweight (BMI P 85–95) and obese BMI≥P 95. The reference utilized was NCHS/CDC 2000 Growth Charts,8 which, although was developed to evaluate the nutritional status of US children, has been endorsed by the Chilean Pediatric Society.9 WC was measured standing with a nonelastic tape that was applied horizontally midway between the lowest rib margin and the iliac crest. TSF was determined with Lange calipers to the nearest 1 mm, three consecutive times in the mid-point of the left arm hanging straight, and averaged.

  • Physical fitness: We applied two health-related tests: the first one assesses flexibility of the lower back by reaching as far as possible from a standing position, while the other one is the endurance 20 m shuttle run test (20 m SRT or Leger and Lambert test),10 which indirectly determines aerobic capacity by running at an increasing speed back and forth a distance of 20 m. The subjects start running at a speed of 8.5 km/h, then increase at various stages in accordance with the pace dictated by a sound signal, which gets progressively faster. Each stage of the test is made of several shuttle runs, but the actual score is the last half-stage fully completed before the subject drops out. This test has been found to be reliable (test–retest r=0.69–0.87).11 The flexibility test was applied to all children, while the endurance run to those children between 4th and 8th grades, because among younger children results were found to be less reliable.12 Application of these tests was recommended by specialists from the Government Sports Promotion Agency, based on their inclusion in the European Test of Physical Fitness (Eurofit).13

Secondary outcome measures

These included a selected dietary assessment, attitudes and behavior towards healthy eating and physical activity

  • Selected dietary assessment questionnaire consisted in a food consumption survey of key food items. The survey was completed by the children from fourth to eighth grade on two different days (a weekday and the corresponding Monday) of the same week, based on consumption of the previous 24 h. A total of 16 items, including healthy foods (fruits, vegetables and dairy), energy-dense snacks (high in fat/sugars processed foods), high-sugar beverages, as well as fast foods, served to assess frequency of food consumption and also to determine if there were differences in consumption between weekday and Sunday. We decided to determine frequency rather than actual intake, after reviewing recent studies in Europe and the USA, which have used this methodology to gather information regarding food consumption in large populations of children.14,15

  • Attitudes and behavior related to healthy eating and physical activity: These were assessed on children from 4th to 8th grade and consisted in a self-registered questionnaire on a day different from that of food consumption. It included questions related to physical activity and healthy eating. We designed specific questions and selected others from research carried out in the USA and Europe based on children's ability to understand them.14,15 A total of 14 questions related to physical activity (attitudes, behavior, perception and barriers). They included the number of ‘active and sedentary’ activities performed after school, engagement and liking of vigorous physical activity, and perceived ability to engage in physical activity. With respect to healthy eating, only questions related to fruits and salads were investigated (liking, willingness to increase consumption and perception if amount consumed was adequate)

Questionnaires were pilot tested before the study was begun in children of similar SES attending summer schools who would enter fourth to eighth grade the following school year. The questionnaires were modified until all questions were understood by most children. We found that about 15% of children who would be entering 4th grade had difficulty in reading and writing. Based on this observation, we were prepared to provide special assistance to those with difficulties during the actual study.

Data collection related to process

  • Kiosks: We monitored the monthly sales of food items bought by the children during the mid-morning recess both in intervention and control schools. The research nutritionist registered all sales for the corresponding recess on a given day near the middle of the month. We then determined the proportion of ‘healthy foods’ sold. The following items were considered healthy: dairy products, fruits, sandwiches, seeds and dry fruits.

  • Parents: Data were collected in November on a sample of parents (one class per grade, from 1st to 8th, around 329 parents) about their general opinion regarding the intervention and their perception of possible changes in diet and/or physical activity observed during the school year.

  • Teachers: They recorded the approximate amount of time spent in implementing each of the programmed activities of the diet/nutrition educational program. Also, they (26 of 40 total) provided through a structured questionnaire (14 questions) and one open question their opinion regarding the educational program, the type of activities proposed, the difficulty in applying the program as planned and the support given by the nutritionist.

This manuscript will solely present results of the primary outcome variables in intervention and control schools. These include anthropometric indices (BMI, BMI Z score, TSF, WC) and physical fitness indices (20 m SRT scores and flexibility) measured on all the children before and after the 6-month intervention period, and changes observed according to sex, school and baseline status. Secondary outcomes and process indicators will be informed when they contribute to a better understanding of the effects on primary outcomes. These will be reported in greater detail elsewhere.

Statistical analysis

Although the number of schools included was defined a priori, we assessed power and significance that resulted from a total sample size of 3086 subjects with a ratio of two experimental subjects per one control. We based the power estimation considering an effect of at least 0.35 BMI units (approximately 0.1 s.d.) difference between the two groups with an alpha of 0.05; the post hoc power estimated was 0.8. This calculation assumed that BMI in the intervention schools would remain unchanged, while that of control schools would increase by approximately 0.3 BMI units, which is the expected increase in children of this age over 6 months. The s.d. to calculate the sample size in both groups was obtained from existing information on children of similar age in Chile and is consistent with other international data.16,17

t-Tests for independent samples and nonparametric Mann–Whitney tests were conducted to assure comparability of anthropometric and fitness measurements between groups at baseline. The effect of the program was analyzed using a mixed model analysis of covariance, adjusting for gender and baseline value, estimating the relative changes observed in the two groups, over the 6-month study period. The two-way model tested the effect of time, treatment and interaction. As TSF measures and the 20 m SRT scores (stages) were not normally distributed, the analyses for those variables were performed using natural log transformed values. We examined height/age Z-score (HAZ) to see if there were differences in this parameter at baseline and explore change over time, using the procedure mentioned above. In addition, we also assessed changes over time within each group (differences between schools) in response to the intervention. The comparative analysis of changes was adjusted by school effect where appropriate. All analyses were performed using the SAS statistical package (SAS release 8.2, 2002, SAS Institute, Canada).


The baseline sample included 3577 children from both intervention and control schools. Baseline and follow-up values for anthropometry were collected in 3086, that is, 86.3% of the total sample. The main reason for missing data was child's absence from school on days of measurements. Fitness variables were taken on subsamples of total population as detailed under Methods. Table 1 shows the baseline characteristics of 2141 children in the intervention and 945 in the comparison schools. The average age was 10.6 y for both sexes; the proportion of boys was slightly higher in both groups. No significant differences between intervention and control schools were found in mean values for weight, height, BMI and TSF. The proportion of obese children was significantly higher in the experimental group; differences were also noted for mean values in BMI Z-score, BMI percentile and WC. Concordantly with the higher obesity prevalence, physical fitness indices were worse (lower values) in the experimental schools. These results are consistent with a nonrandom assignment of schools to experimental and control groups. As explained under Methods, there was a clear bias favoring greater obesity in the experimental schools.

Table 1: Characteristics of the sample at baseline in intervention and control schools (values are mean and (s.d.) unless stated otherwise)

The effect of the intervention over time was evaluated based on changes in the values of anthropometric and physical fitness indices between baseline and follow-up. Table 2 shows these effects in boys. Average BMI remained unchanged in the intervention group over time, while in control schools it increased by 0.3 U, as expected for this age over 6 months. The analysis shows that this change was significantly affected by the interaction between group assignment and time, after adjusting for baseline BMI values. The BMI Z-score declined significantly in the intervention schools, while in control schools it remained unchanged; the group × time interaction was significant (P<0.001). HAZ in boys was significantly lower in control schools at baseline; it remained unchanged over time in intervention schools, while the mean dropped by 0.05 s.d.'s in control schools. TSF dropped nonsignificantly over time in both groups. WC declined in the intervention group by a mean of 0.9 cm, while in controls it increased by the same amount; a significant group effect and a highly significant interaction of group × time were noted (P<0.0001). Both physical fitness parameters improved significantly in the intervention group (P<0.001), while in the control group the 20 m SRT remained unchanged and flexibility declined.

Table 2: Change in anthropometric and physical fitness variables at follow-up in boys from intervention and control schools

For girls (Table 3), BMI and BMI Z scores were not significantly affected by the intervention. HAZ decreased in the intervention schools, while it remained unchanged in control schools; there was a group × time interaction, indicating a greater drop in the intervention group. TSF and WC were not significantly affected by the intervention, although a slight increase was noted in both groups. There was a significant effect of the intervention on physical fitness; both the 20 m SRT scores and flexibility improved significantly in the intervention schools, while they declined in control schools (P&lt;0.001 for each test).

Table 3: Change in anthropometric and physical fitness variables at follow-up in girls from intervention and control schools

In order to examine if the effect of the intervention (delta BMI Z) was homogenous and independent of baseline BMI Z-score, we evaluated this change for obese, overweight, normal weight and underweight. The results of this analysis showed a significant effect of baseline BMI Z in both boys and girls. For combined sexes, the data demonstrated that the obese and overweight dropped by −0.16 and −0.15 BMI Z, respectively, normal children lost 0.07 Z while those underweight gained +0.15 BMI Z units. For the control group, there was no effect in terms of observed BMI Z over the 6 months of study, changes ranged from +0.07 for the underweight, to −0.06 for normals, no change for overweight and +0.03 for the obese. These findings are unique and of key significance for developing countries that must face the challenge of targeting a combination of under- and overnourished population.18

We also explored the effect of age on change in BMI Z in both intervention and control groups. No significant overall effect of age was found when assessing the interaction of age, group and time of measurement (baseline and at follow-up); P-value was 0.28 for boys and 0.06 for girls. However, if both control and intervention groups were examined jointly, for boys, those older than 12 y of age had a greater decline in BMI Z, suggesting an independent effect of pubertal maturation. In the case of girls for both groups combined, the nearly significant interaction showed a greater drop in BMI Z in those younger than 12 y of age. The age effect in this case is difficult to interpret; again, this is an effect which is independent of the intervention.

School effects on changes in main outcomes between baseline and follow-up were also analyzed using the mixed model of covariance. The mean BMI Z scores for boys and girls in each of the three intervention schools compared to mean±s.e. for control schools combined are shown in Figure 1. For boys, significant differences among intervention schools were noted at baseline; BMI Z scores declined significantly over time in each of them, while in control schools these remained unchanged. No group × time interaction was observed. For girls, we also noted differences at baseline between intervention schools, but, in contrast to boys, mean BMI Z scores were not affected by the intervention. There was a significant individual time and group effect, but no group × time interaction.

Figure 1
Figure 1

Comparison of BMI Z scores between baseline and follow up in intervention and control schools according to gender.

The mean 20 m SRT scores for boys and girls in each of the three intervention schools compared to mean±s.e. for control schools combined are shown in Figure 2. For boys, we observed a small but significant difference between schools at baseline (P=0.04). In all intervention schools, there was a major improvement in this score at follow-up; the group × time interaction was highly significant, while in controls this score remained unchanged. For girls, in contrast to boys, no differences between intervention schools were noted at baseline, but at follow- up, an important improvement occurred; this group × time interaction was highly ignificant. In this case, control schools had a significant decline in 20 m SRT scores at follow-up.

Figure 2
Figure 2

Comparison of the 20m Shuttle Run Test scores between baseline and follow up in intervention and control schools according to gender.

We provide preliminary results on process indicators related to program implementation in order to illustrate their potential impact on the main outcomes reported in this paper. The implementation of the nutrition education program did not occur fully as planned. In school B, teachers applied 100% of the planned educational activities vs 85% in school C and 80% in school A. Most teachers underwent a short training and only about half of them were initially willing to take the responsibility for this activity. Later on, the rest accepted the challenge.

The intervention had absolutely no effect in modifying the pattern of sales of healthy foods by the kiosks. At baseline, 3.5% of the total number of food items sold over a recess in intervention schools and 0.34% in control schools were considered healthy as defined under Methods. At the follow-up evaluation, the proportion of healthy foods sold was 2.6 and 0%, respectively.

The provision of extra physical activity time was successfully implemented in all schools, because it was incorporated into the curriculum. When no regular teacher was available for this activity, the research PE teacher assumed this role. Active recess was also implemented successfully in all schools. In this study, we could only assess subjectively the influence it had on overall physical activity based on observations by the research team and school teachers assigned to supervise the children during recess. All opinions coincided in concluding that most children enjoyed the music or were more active because they had more sports equipment available. Interestingly, the number of injuries related to accidents or fights decreased substantially, from 168 to 68; overall decrease for all schools during the 3-month observation was 60%.

Parental involvement based on participation in scheduled meetings was much lower in Santiago, approximately 50%, relative to Curico and Casablanca (smaller cities), where it was around 90%. A total of 83% of parents who provided their general opinion on the intervention reported positive changes in their children both in dietary intake and/or physical activity; 66% said they themselves incorporated some aspects of what was discussed at meetings in their own lifestyle.


In Chile, the prevalence of obesity among children has risen sharply over the past two decades. In 6-y olds, it increased from 7% in 1987 to 17.2% (weight for height >2 s.d. NCHS) in the year 2000.19 Sedentary behavior among children is also increasingly prevalent. A study conducted recently in a low-income neighborhood of Santiago showed that 10-y-old children during a regular school day, on average, sleep 11.6 h, carry out light activities during 8 h, watch television 3.2 h and spend only 1.2 h in moderate-to-vigorous physical activities.20 In addition, dietary patterns, determined on similar children as the ones included in our study, showed increased consumption of energy-dense foods (snacks high in fat and sugars) and low consumption of energy-dilute foods (fruits and vegetables).17,21

The objective of this study was to evaluate in a controlled manner the effects of a diet/nutrition education and physical activity intervention on obesity and fitness among school-aged children. To assess effectiveness (impact evaluation), we chose simple variables that have been demonstrated to detect changes in body composition and physical activity over short-term periods. The anthropometric measurements that we included in our study, BMI, TSF, have been shown to predict total fat content (measured against results from underwater weighing or DEXA) in children and adolescents.22,23 BMI, although reliably measured, changes significantly with age and sex, so BMI values were also expressed as Z scores. TSF and WC are both simple measurements: TSF relates to total body fat, a potential risk for cardiovascular diseases,24 while WC is associated with the distribution of abdominal fat23 and consequently to the occurrence of the metabolic syndrome. It can thus provide insights into the rate of visceral fat accumulation during childhood and adolescence.25,26 Pubertal stage has an important effect on fat accumulation and distribution; BMI for age assumes equal maturation at a given age, thus neglects this source of variance.27 We were unable to collect data on sexual maturation, due to concern over the children's privacy expressed by the educational authorities. We thus recognize this issue as a source of uncontrolled variance in the anthropometric results.

Our data indicate that at baseline, the obese (all schools) were in fact 0.7 y younger, indicating a possible systematic effect of earlier maturation on obesity prevalence. However, this effect was not of major significance in the results of our study, since the age distributions of intervened and control groups were similar within each sex category.

Field-based fitness tests for children include measures of cardiorespiratory endurance, muscular strength and endurance and flexibility.28 Of these measures, we were only able to perform two, because of time and personnel constraints. We chose cardiorespiratory fitness and flexibility, because these have been widely used in evaluations performed on Chilean school children (published as reports). Cardiorespiratory fitness was determined indirectly according to children's performance on the endurance 20 m SRT through maximum aerobic power; the higher the score the better the cardiovascular function. This field test is recommended for large groups of children since it is reliable, valid, noninvasive and requires limited facilities.10,29

The results showed that in boys, there was a significant positive effect on all adiposity-related indices (except TSF) at follow-up. In contrast, for girls, no effect was noted on adiposity. The effect in boys was observed as crude BMI, BMI Z scores (the effect was stronger when expressed as Z-score, as revealed by the higher F-value) and WC. The control group behaved as expected, increasing BMI by 0.3 U over 6 months, while the experimental group remained unchanged. Concordantly, BMI Z-scores dropped in the intervened group remaining unchanged in the controls. The lack of effect on TSF values may be due to the small change observed, but most probably to the high variability in values, typical of children undergoing pubertal changes. The intra- and interobserver measurement error in TSF was carefully controlled prior to starting the study, so this factor is unlikely to explain the high variability.

The effect of the intervention on each school as determined by BMI Z-scores in boys (Figure 1) was evident in all intervention schools despite the marked differences in baseline values. In girls, values were also different at baseline, but in, contrast, remained practically unchanged at follow-up. The anticipated effect of mean BMI change between baseline and follow-up (+0.3 kg/m2) on adiposity, assuming that this weight (about 600 g) was entirely fat tissue, would correspond to a change in energy balance of around −26 kcal/day. This change in energy balance could be due to increased physical activity and/or decreased energy intake. The methods to assess dietary intake used in our study do not allow for a quantitative estimation of differences in energy intakes of this magnitude; even the best methods would not pick up such a small difference.30

It is notable that for the intervention group, the change in BMI Z-score for the obese compared to those underweight was in the opposite direction, while the underweight gained BMI Z, the obese and overweight lost weight relative to size. This is a highly desirable outcome and reflects the need to examine the differential effect within the intervened; with regard to the control group, there were no changes in BMI Z. This in fact suggests that changes observed in the intervention group are not due to a regression to the mean phenomenon.

All intervention schools exhibited major gains in 20 m SRT scores. At baseline, these schools revealed differences (small) for boys only; mean values ranged from 3.65 to 3.76. Boys from school A improved on average 17%, while in schools B and C they improved 50 and 40%, respectively. In girls, no school effect was observed; as in boys, school B exhibited the greatest improvement, in this case 41%. This is probably due to the fact that school B was substantially smaller; the research PE teacher (who was highly motivated) was directly in charge of most of the additional PE classes. In the other two schools, only approximately 1/3 of the classes were conducted by the research PE teacher. In examining the differences in response to the intervention among schools (Figures 1 and 2), we observe that BMI Z-score behaved similarly in all three intervention schools in boys. In contrast, the change in 20 m SRT was different between schools despite generalized improvement.

The degree of implementation differed among schools; the smaller the enrollment, the better the degree of adherence to the program. We consider this to be due in part to individual motivation by teachers and also to the time devoted by the professionals hired by us to support the process. It is not surprising that results differ according to school. It has been established that a key issue in the ability to deliver a program successfully is that the components and implementation procedures have to harmonize science with what different people involved propose to implement and recipients are willing to accept. As reported by Baranowski et al,31 if for example, teachers do not implement the experimental curricula with substantial fidelity, desired changes may have limited effect. Additionally, public school teachers in general have a very limited knowledge in nutrition,6 and training time was very short.

The additional PE classes were sports oriented and well liked by the children according to the report of 44 teachers from the three intervention schools (75% answered the questionnaire). In fact, 80% of the teachers thought that the additional PE classes should be permanently a part of the regular school curricula. In addition, active recess periods may have contributed to the overall increase in physical activity level. In general, children are not active in recess: observational studies carried out in elementary schools in the USA32,33 found students to be physically active for about 50–60% of the total recess time. In our schools the situation is similar, as documented by Gattas et al;34 therefore, if recess time is more active, total daily physical activity among children should increase.31 The major impact of this study was on physical fitness: both the 20 m SRT scores and flexibility improved significantly in the intervention group in both sexes, while it remained unchanged or even declined in the control group. Boys responded more effectively to the physical activity intervention (determined as cardiorespiratory fitness): boys improved by 35% while girls by 26%. We can hypothesize that fitness in this case relates to increased physical activity and thus possibly increased energy expenditure. It has been shown that girls are less active than boys,35,36,37 so probably boys engaged in more vigorous activity during the additional PE classes, and this was associated not only with an improvement in physical fitness but also with a reduction in adiposity. Our data on food frequency questionnaire will serve to explore if responders in terms of BMI changed consumption patterns of specific food items more frequently and thus a combined effect of changes in physical activity and food intake may offer a better explanation for the gender differential on BMI response.

The results of the ‘healthier’ kiosks intervention demonstrated that the availability of foods of low nutritional value remained unchanged. From a nutritional standpoint, the nature of these foods is of considerable concern; they included carbonated soft drinks, cookies, chocolates, candies and chips. Studies have shown that most low-income children buy these foods at school.38 Unfortunately, there is no regulation restricting the sale of these items, and schools need the revenues to pay some basic services. One successful strategy was the contest called ‘Healthy Snack’, which motivated children to eat a healthy food during recess. Gifts and prizes have been used as incentives for adopting health-enhancing behaviors. For example, there have been interventions to promote the selection of low-fat milk, and to increase physical activity.39

The main limitation of this study is that it is a nonrandomized trial, because school assignment to the intervention group was intentionally made by the local educational authorities. This resulted in higher obesity rates in intervention schools and conversely, better fitness parameters in control schools. In fact one could consider that this would bias the results toward a no effect: in contrast, we found a significant impact of the intervention. Having a control group allowed us to assess the influence of confounding variables, such as age, sex, type of school and baseline values. Dealing with confounders is very important in the statistical treatment of results.40

School-based preventive strategies have not been demonstrated to be generally effective at impacting obesity rates. Resnicow and Robinson41 reviewed 16 major school-based cardiovascular prevention trials published from 1985 to 1995, analyzing results through semiquantitative metaevaluations. Positive effects were observed more frequently for smoking and cognitive aspects, while the lowest effects were observed for adiposity measures. Jacob42 also reviewed school-based programs that target obesity, most of them with a time frame over 2 y. She concluded that in almost all studies, knowledge and attitudes improved and more consistent data resulted from improvement of physical activity intensity during PE classes as well as reduction of fat content of school meals. Robinson43 on the other hand reports that ‘behavior change’ interventions instead of ‘health education’ interventions seem to be more successful in achieving a reduction in BMI, skinfolds and aerobic capacity. Results from this controlled study should be interpreted in relation to comparable short-term interventions. Campbell et al5 assessed the effectiveness of long- and short-term educational, health promotion and/or other type of intervention that focused on diet, physical activity or other aspect related to lifestyle and were designed to prevent obesity in childhood. The review included only data from randomized and nonrandomized trials with concurrent control groups published from 1985 till 1997. Although the four short-term studies that met their inclusion criteria are not strictly comparable to our study, because not only were they randomized controlled studies, but two of them were only physical activity interventions,44,45 one aimed at reducing TV time while the last one by Stolley and Fitzgibbon46 sought to compare a dietary education vs a physical activity intervention, it is interesting to acknowledge the outcomes. Flores et al found that in a sample of 10–13-y old African-American and Hispanic children, BMI declined significantly in girls while fitness improved in both genders, after 12 weeks of 150 min of dance per week. The study by Mo-Suwan, which included preschool children, and consisted in delivering a specific daily regimen of exercise (20 min of walking plus 20 min of aerobics) for 29.6 weeks, also found a reduction of the obesity prevalence that nearly reached significance. ‘The Active program promoting lifestyle in schools (APPLES)’47 implemented in some schools in the UK is comparable to our study; it uses a population health promotion approach targeting the whole school community including parents and teachers, over 12 months. The main difference is the attempt to modify school meals. Sahota et al,47 who implemented the APPLES program in primary schools in Leeds, reported improvement at the school level, but no changes in BMI in the children from intervention schools.

This study reveals important pitfalls in achieving uniform efficacy. Teachers play a key role in successful implementation; parental involvement was shown to be quite difficult to achieve and potentially critical to influence the home environment. Our kiosk intervention failed in the absence of clear incentives to provide healthier choices or regulations to limit the availability of energy-dense snacks and sugar-rich drinks. On the positive side, we were able to achieve significant improvements across genders and schools in physical fitness; the decreased adiposity found in boys is most likely due to their higher compliance with the PA intervention and greater intensity of their activity leading to a small but measurable negative energy balance. In addition, this intervention produced a desirable outcome relative to baseline nutritional status. Interventions need to be tested over a longer time period and encompass the multiple factors that contribute to the rising obesity prevalence. As demonstrated by this study, this is easier said than done.


  1. 1.

    , , , . Nutrition transition in Latin America: the case of Chile. Nutr Rev 2001; 59: 170–176.

  2. 2.

    . An overview on the nutritional transition and its health implications: the Bellagio meeting. Public Health Nutr 2002; 5: 93–103.

  3. 3.

    , , , , . Nutrition transition in Chile: determinants and consequences. Public Health Nutr 2002; 5: 123–128.

  4. 4.

    , . Health promotion in Chile. Rev Chil Nutr 2002; 50: 164–173.

  5. 5.

    , , , , . Interventions for preventing obesity in children (Cochrane Review). In Cochrane Library, issue 1. Update Software: Oxford; 2003.

  6. 6.

    , , , , . Nutrition education in Chilean primary schools. Food, Nutr Agric 2003; 33: 64–69.

  7. 7.

    CALC Program. CALC Web Page ().

  8. 8.

    CDC/NCHS. CDC Growth charts: United States, Posted May 30, 2000 on Internet 2000.

  9. 9.

    Chilean Ministry of Health. Normas de evaluación nutricional del niño de 6 a 18 años 2003.

  10. 10.

    , . A maximal multistage 20 m shuttle run test to predict VO2max. Eur J Appl Physiol 1982; 49: 1–5.

  11. 11.

    , , , . The multistage 20 metre shuttle run test for aerobic fitness. J Sports Sci 1988; 6: 93–101.

  12. 12.

    , , . Relative efficiency and predicted VO2 max in children. Med Sci Sports Exe 1983; 15: 143–147.

  13. 13.

    Committee of Experts on Sports Research. EUROFIT. Edigraf, Editoriale Grafica: Rome 1988.

  14. 14.

    WHO. Health Behaviour in School-Aged Children. Research Protocol for the 1997–98 Survey. Department of Community Health Sciences, The University of Edinburgh, Edinburgh, Scotland; 2000.

  15. 15.

    CDC. US Youth Risk Behavior Surveillance System, Information on the Web Page ().

  16. 16.

    . Reducing children's television viewing to prevent obesity: a randomized controlled trial. JAMA 1999; 282: 1561–1567.

  17. 17.

    , , , . Nutritional status, food consumption and physical activity in Chilean school children: a descriptive study. submitted for publication.

  18. 18.

    , , , . Overweight and underweight coexist within households in Brazil, China and Russia. J Nutr 2000; 230: 2965–2971.

  19. 19.

    , , , . Trends in overweight and obesity in Chilean children: comparison of three definitions to determine prevalence. Eur J Clin Nutr 2002; 56: 200–204.

  20. 20.

    , , , . Validation and implementation of questionnaires to evaluate school-based obesity prevention educational interventions. Rev Chil Pediatr 2001; 72: 308–318.

  21. 21.

    , , , , . Food consumption in Chilean school children: relationship with food guides and pyramid. Rev Chil Nutr 2001; 28: 422–428.

  22. 22.

    , , , , , . Body mass index, triceps skinfold and waist cicumference in screening for adiposity in male children and adolescents. Acta Pediatr 2001; 90: 387–392.

  23. 23.

    , , , , . Predicting body composition from anthropometry in pre-adolescent children. Int J Obes Relat Metab Disorder 1999; 23: 253–259.

  24. 24.

    , , . Parameters of childhood obesity and their relationship to cardiovascular risk factors in healthy prepubescent children. IJOB 2001; 25: 830–837.

  25. 25.

    , , . Waist measurement correlates to a potentially artherogenic lipoprotein profile in obese 10–12 year-old children. Acta Pediatr 1994; 83: 941–945.

  26. 26.

    , . Waist percentiles: a simple test for atherogenic disease? Acta Pediatr 1996; 85: 1368–1369.

  27. 27.

    , , . Prevalence of obesity in adolescents and the impact of sexual maturation on BMI in obese adolescents. Int J Adolesc Med Health 2002; 14: 61–65.

  28. 28.

    , , . Status of field-based fitness testing in children and youth. Prev Med 2000; 31: 577–585.

  29. 29.

    , , . Gender differences in physical activity and physical fitness in young children in Crete. J Sports Med Phys Fitness 1999; 39: 24–30.

  30. 30.

    , . Markers of the validity of reported energy intake. J Nutr 2003; 133: 895S–920S.

  31. 31.

    , , , , . Observations on physical activity in physical locations: age, gender, ethnicity, and month effects. Res Q Exerc Sport 1993; 64: 127–133.

  32. 32.

    , , , , , , , . Physical activity levels and prompts in young children at recess: a two-year study of a bi-ethnic sample. Res Q Exerc Sport 1997; 68: 195–202.

  33. 33.

    . Children at play: behavior of children at recess. JOPERD 1989; 60: 21–24.

  34. 34.

    , , , . Physical activity in normal and low height for age school children. Rev Chil Pediatr 1996; 67: 212–218.

  35. 35.

    , , , , . Effects of a curriculum and inservice program on the quantity and quality of elementary physical education classes. Res Q Exer Sport 1993; 64: 178–187.

  36. 36.

    , , , , , , . Compliance with physical activity guidelines: prevalence in population of children and youth. Ann Epidemiol 2002; 12: 303–308.

  37. 37.

    , , , , , . Gender and development differences in exercise beliefs among youth and prediction of their exercise behavior. J Sch Health 1995; 65: 213–219.

  38. 38.

    , , , . Television advertising and food preferences of school age children of Santiago, Chile. Rev Med Chil 1999; 127: 791–799.

  39. 39.

    , , , . Using the school environment to promote physical activity and healthy eating. Prev Med 2000; 31: S121–S137.

  40. 40.

    , , . Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. Int J Epidemiol 1999; 28: 10–18.

  41. 41.

    , . School-based cardiovascular disease prevention studies: review and synthesis. Ann Epidemiol 1997; S7: S14–S31.

  42. 42.

    . School programs. In: Chen C, Dietz W (eds). Obesity in childhood and adolescence. Nutrition Workshop Series. Pediatric Program. Vol. 49. Lippincott: Philadelphia; 2002. 257–272.

  43. 43.

    . Obesity Prevention. In: Chen C, Dietz W (eds). Obesity in childhood and adolescence. Nutrition Workshop Series. Pediatric Program. Vol 49. Lippincott: Philadelphia; 2002. 245–256.

  44. 44.

    . Dance for health: improving fitness in African American and Hispanic adolescents. Public Health Rep 1995; 110: 189–193.

  45. 45.

    , , , . Effects of a controlled trial of a school-based exercise program on the obesity indices of preschool children. Am J Clin Nutr 1998; 68: 1006–1111.

  46. 46.

    , . Effects of an obesity prevention program on the eating behavior of African American mothers and daughters. Health Educ Behav 1997; 24: 152–164.

  47. 47.

    , , , , , . Randomized controlled trial of primary school based intervention to reduce risk factors for obesity. BMJ 2001; 323: 1–5.

Download references


This study was supported by the Chilean Ministry of Education, Chile Deportes (Government Sports Promotion Agency) and an unrestricted grant from Córpora Tresmontes.

Author information


  1. Institute of Nutrition and Food Technology (INTA), University of Chile, Santiago, Casilla 138-11, Santiago, Chile

    • J Kain
    • , R Uauy
    • , Albala
    • , F Vio
    • , R Cerda
    •  & B Leyton
  2. Public Health Nutrition, London School of Hygiene and Tropical Medicine, London, UK

    • R Uauy


  1. Search for J Kain in:

  2. Search for R Uauy in:

  3. Search for Albala in:

  4. Search for F Vio in:

  5. Search for R Cerda in:

  6. Search for B Leyton in:

Corresponding author

Correspondence to J Kain.

About this article

Publication history






Further reading