| Methods |
Study design: cluster‐randomized controlled trial
Study grouping: parallel group
Study aim: "... to evaluate the effects of a middle school physical activity and healthy eating intervention, including an environmental and computer‐tailored component, and to investigate the effects of parental involvement" (quote)
Study period: baseline: September 2003; final assessment: May to June 2005
Total number of arms: 3
Description of intervention arms: 1. School‐based intervention with parental support (child + caregiver); 2. School‐based intervention (child only); 3. No intervention control (not eligible)
Number of clusters per arm: 5
Average cluster size: 189.33 children
Sample size justification and outcome used: to detect a 0.3‐change in body mass index (BMI) and a 10‐minutes‐per‐day change in total physical activity with 80% power and an alpha level of 0.05, 300 children per group were needed. Study authors sought larger samples to account for possible dropout and to support subgroup analyses
Unit of allocation: schools
Missing data handling: to account for missing data on physical activity outcomes, "intention‐to‐treat analyses were conducted by carrying the last observation forward" (quote). However, the number analyzed differs from the baseline sample, and the reason why is not explained. For BMI and dietary intake analyses, study authors did not report how missing data were handled. In data provided by the study authors, the distinct sample sizes used to assess baseline to 1‐year outcomes vs baseline to 2‐year outcomes suggest that data were analyzed based on matched pairs
Reported limitations: 1. Randomization at the school level may have contributed to clustering of outcomes (but this was considered in analyses); 2. Considerable gender differences were evident across study arms (accounted for in analyses); 3. Intervention period was too short to observe sustained behavior change; 4. Accelerometers were used only in a subsample of seventh graders; 5. Quality of implementation and parental involvement was not objectively evaluated; 6. It was not possible to isolate the effects of specific intervention components; 7. After 2 school years, there was a relatively high percentage of dropouts (25%); 8. Self‐reported diet and physical activity measures are subject to reporting errors
Randomization ratio and stratification: 1:1:1; stratification not reported
Participant compensation or incentives: not reported |
| Participants |
Baseline characteristics
Child + caregiver arm (intervention group)
Female (PROGRESS‐Plus): n (%): 479/1194 (40.1) Age in years (PROGRESS‐Plus): mean (SD): 13.04 (0.79) Race/ethnicity/language/culture (PROGRESS‐Plus): not reported Place of residence (PROGRESS‐Plus): not reported Caregiver education (PROGRESS‐Plus): not reported Religion (PROGRESS‐Plus): not reported Household income/socioeconomic status (PROGRESS‐Plus): lower socioeconomic status, n (%): 812/1194 (68.0) Social capital (PROGRESS‐Plus): not reported Caregiver work hours and other characteristics that may indicate disadvantage (PROGRESS‐Plus): not reported Disability (PROGRESS‐Plus): not reported Sexual orientation (PROGRESS‐Plus): not reported Child weight status: BMI Z score: mean (SD) 0.14 (1.10) Child diet: % energy from fat, mean (SD): 38.7 (16.3); fruit (pieces/week), mean (SD): 5.3 (5.3); SSB (glasses/d), mean (SD): 3.1 (2.4); water (glasses/d): 3.4 (2.7) Child physical activity: total physical activity (min/d), mean (SD): 94.8 (53.9) Caregiver weight status: not reported Caregiver diet: not reported Caregiver physical activity: not reported Caregiver civil status (PROGRESS‐Plus): not reported
Child‐only arm (control group)
Female (PROGRESS‐Plus): n (%): 142/911(15.6) Age in years (PROGRESS‐Plus): mean (SD): 13.24 (0.87) Race/ethnicity/language/culture (PROGRESS‐Plus): not reported Place of residence (PROGRESS‐Plus): not reported Caregiver education (PROGRESS‐Plus): not reported Religion (PROGRESS‐Plus): not reported Household income/socioeconomic status (PROGRESS‐Plus): lower socioeconomic status, n (%): 719/991 (78.9) Social capital (PROGRESS‐Plus): not reported Caregiver work hours and other characteristics that may indicate disadvantage (PROGRESS‐Plus): not reported Disability (PROGRESS‐Plus): not reported Sexual orientation (PROGRESS‐Plus): not reported Child weight status: BMI Z score: mean (SD) 0.13 (1.03) Child diet: % energy from fat, mean (SD): 43.7 (18.1); fruit (pieces/week), mean (SD): 4.6 (5.0); SSB (glasses/d), mean (SD): 3.5 (2.5); water (glasses/d): 3.1 (2.7) Child physical activity: total physical activity (min/d), mean (SD): 100.9 (58.7) Caregiver weight status: not reported Caregiver diet: not reported Caregiver physical activity: not reported Caregiver civil status (PROGRESS‐Plus): not reported
Recruitment methods: to obtain a sample of 15 of the 65 Flemish technical and vocational schools in West Flanders, Belgium, the research team contacted principals of 23 of the schools by phone; 8 declined participation. Following phone contact, principals were sent more information by mail, and a meeting with the research team was arranged. Parents of all seventh and eighth grade students in the 15 participating schools received an informed consent form to provide authorization for their child to participate in measurements
Inclusion criteria: cluster: random sample of 15 of the 65 schools with technical and vocational education in West Flanders, Belgium; participants: seventh or eighth grade students and caregivers
Exclusion criteria: not reported
Age of participating children at baseline: seventh to eighth grade (estimated to be 11 to 14 years old)
Total number randomized by relevant group: total across all study arms: n = 2840; child + caregiver arm: n = 1194, child‐only arm: n = 911
Baseline imbalances between relevant groups: at baseline, no significant differences were reported between study arms
Total number analyzed by relevant group: the number analyzed differs for each measure. Data provided by the corresponding author: total kcal at 9‐month assessment (child + caregiver arm: n = 1070, child‐only arm: n = 678) and at end of intervention (child + caregiver arm: n = 902, child‐only arm: n = 635); % kcal from fat at 9‐month assessment (child + caregiver arm: n = 970, child‐only arm: n = 563) and at end of intervention (child + caregiver arm: n = 738, child‐only arm: n = 514); fruit intake at 9‐month assessment (child + caregiver arm: n = 1064, child‐only arm: n = 691) and at end of intervention (child + caregiver arm: n = 902, child‐only arm: n = 580); soft drink intake at 9‐month assessment (child + caregiver arm: n = 1087, child‐only arm: n = 660) and at end of intervention (child + caregiver arm: n = 934, child‐only arm: n = 595); water intake at 9‐month assessment (child + caregiver arm: n = 1086, child‐only arm: n = 660) and at end of intervention (child + caregiver arm: n = 932, child‐only arm: n = 595); total physical activity at 9‐month assessment (child + caregiver arm: n = 1081, child‐only arm: n = 676) and at end of intervention (child + caregiver arm: n = 883, child‐only arm: n = 632); MVPA and light physical activity at 9‐month assessment (child + caregiver arm: n = 70, child‐only arm: n = 52) and at end of intervention (not reported or provided for relevant study arms); BMI Z score at 9‐month assessment (child + caregiver arm: n = 1127, child‐only arm: n = 882) and at end of intervention (child + caregiver arm: n = 980, child‐only arm: n = 768); prevalence of overweight/obesity at 9‐month assessment (child + caregiver arm: n = 1129, child‐only arm: n = 888) and at end of intervention (child + caregiver arm: n = 1080, child‐only arm: n = 907)
Attrition by relevant group: attrition rates were calculated by dividing the number of children missing data at the assessment time point by the number of children randomized with consent. Rates differed by outcomes and are based on unpublished information provided by the study authors. Dietary outcomes, 9‐month assessment: child + caregiver arm: 11.6% (139/1194), child‐only arm: 24.8% (226/911), end of intervention: not reported or provided by relevant study arms; self‐reported physical activity, 9‐month assessment: child + caregiver arm: 9.4% (70/1194), child‐only arm: 7.4% (68/911), end of intervention: not reported or provided by relevant study arms; accelerometers: not reported or provided by relevant study arms; BMI, 9‐month assessment: child + caregiver arm: 7.0% (78/1194), child‐only arm: 8.0% (73/911), end of intervention: child + caregiver arm: 18.7% (223/1194), child‐only arm: 20.0% (182/911)
Description of sample for baseline characteristics reported above: for female, age, socioeconomic status: children with parental consent and baseline data (child + caregiver arm: n = 1194; child‐only arm: n = 911); child diet: (child + caregiver arm: n = 1055; child‐only arm: n = 685) (Haerens 2007a); total physical activity: (child + caregiver arm: n = 1124; child‐only arm: n = 843) (Haerens 2007b) |
| Interventions |
Intervention characteristics
Child + caregiver arm (intervention group)
Brief name/description (TIDieR #1): school‐based physical activity and healthy eating program with environmental and computer‐tailored components plus caregiver support Focus of intervention: physical activity and diet Behavior change techniques: in addition to the child‐only arm, the following techniques were applied separately or differently in the child + caregiver arm: "feedback and monitoring," "social support," "shaping knowledge," "natural consequences" Why: rationale, theory, or goal (TIDieR #2): in addition to the child‐only arm, study authors noted a gap in evidence regarding whether caregiver involvement increases intervention effects. As with the child‐only arm, the adult version of the computer‐tailored intervention was based on the transtheoretical model and the theory of planned behavior How, where, and when and how much (TIDieR #6 to 8): in addition to the child‐only intervention, schools in the child + caregiver arm set up an interactive meeting for caregivers and sent information on healthy eating and physical activity to each caregiver’s home address 3 times a year in school papers and newsletters. Caregivers also received a CD with the adult computer‐tailored intervention for fat intake to complete at home Who: providers (TIDieR #5): same as child‐only arm Economic variables and resources required for replication: not reported Strategies to address disadvantage: not reported Subgroups: same as child‐only arm Assessment time points: baseline, 9 months, 21 months (end of intervention) Co‐interventions: not reported What: materials and procedures (TIDieR #3 to 4): in addition to materials and procedures provided to the child‐only arm, caregivers were invited to an interactive meeting on healthy eating, physical activity, and health, and were mailed information on healthy food and physical activity 3 times each year in school papers and newsletters. Caregivers were given a CD with the adult version of the computer‐tailored intervention for fat intake and physical activity to complete at home. Through a folder, they were informed that their child had completed the same program at school and were asked to discuss the results together and to support their child in making healthy changes Tailoring (TIDieR #9): in addition to the child‐only arm, caregivers were given a copy of the adult version of the computer‐tailored intervention for fat intake and physical activity, which generated personalized output. They were told that their child had completed a parallel activity in school and were asked to discuss the feedback together Modifications (TIDieR #10): not reported How well: planned and actual (TIDieR #11 to 12): same as child‐only arm Sensitivity analyses: same as child‐only arm
Child‐only arm (control group)
Brief name/description (TIDieR #1): school‐based physical activity and healthy eating program with environmental and computer‐tailored components Focus of intervention: physical activity and diet Behavior change techniques: "feedback and monitoring," "shaping knowledge," "natural consequences," "comparison of behavior," "repetition and substitution," "antecedents" Why: rationale, theory, or goal (TIDieR #2): prior research suggests that targeting the school environment is one promising strategy for promoting diet and physical activity behavior change. However, study authors argued that greater effects could potentially be achieved by combining general school interventions with personalized interventions. At the time of this study, computer‐tailored interventions in health behavior change were emerging and showed promise. This intervention sought to combine a general environmental intervention with a personal intervention delivered in the classroom. The feedback provided by the computer‐tailored interventions was based on the transtheoretical model and the theory of planned behavior How, where, and when and how much (TIDieR #6 to 8): for the physical activity environmental intervention, schools organized an average of 4.7 hours of extra physical activities per week. Sports materials provided as part of the intervention were made available in schools during breaks, at noon, and/or during after‐school hours. Over the 2 years, participants spent 4 class hours on the individualized physical activity component. Participants took a physical fitness test (10 minutes cycling on a computerized cycle ergometer) at the beginning of the second intervention year. Once each school year, children also completed the 1‐hour computer‐tailored physical activity intervention. For the food intervention, schools were asked to make fruit and water available for free or at a very low cost. Over the 2 school years, children spent a total of 2 class hours (1 hour per year) on the individual computer‐tailored intervention for fat and fruit intake Who: providers (TIDieR #5): the intervention was implemented by a working group of school staff. Each school’s work group met with the intervention at staff at the beginning of each school year and then once every 3 months, for a total of 8 contact hours over 2 years. In the first intervention year, a researcher met with the working group every 3 months to plan for future actions. In the second year, the working groups continued without further guidance by the research staff Economic variables and resources required for replication: not reported Strategies to address disadvantage: not reported Subgroups: all analyses included an interaction term for gender. No significant intervention effects were found among boys after 1 or 2 years of intervention. Among girls, a significantly lower increase in BMI Z score (P = 0.05) was observed in the child + caregiver arm when compared with the child‐only arm after 2 years of intervention. Outcome measures were also assessed by a school's level of implementation in line with the process evaluation Assessment time points: baseline, 9 months, 21 months (end of intervention) Co‐interventions: not reported What: materials and procedures (TIDieR #3 to 4): at each school, the work group received the intervention manual and educational materials and met to discuss intervention implementation and plan actions. The physical activity environmental intervention focused on increasing children’s MVPA to at least 60 minutes per day and involved creating more opportunities (including non‐competitive opportunities) for children to be active. Sports materials (e.g. ropes, balls, frisbees) were provided to schools in an intervention box. The individualized physical activity intervention involved a fitness test (using a computerized cycle ergometer) with feedback on fitness level and how to improve it, as well as a computer‐tailored intervention, which consisted of an online questionnaire that generated personalized output feedback. Based on the output, children were asked to complete a task concerning the advice received. The nutrition intervention targeted intake of fruit, soft drinks, water, and fat. The fruit component involved asking schools to make fruit available for free or at very low cost with the aim of increasing intake to at least 2 pieces per day. The beverage component focused on promoting drinking water (at least 1.5 liters per day) instead of soft drinks and involved offering water for free at drinking fountains and pricing it lower than soft drinks in shops and from vending machines. Children also received information about the health benefits of choosing fruit over other snacks and water over soft drinks through folders and posters. The computer‐tailored nutrition intervention consisted of a questionnaire to assess fat and fruit intake. As with the computer‐tailored physical activity intervention, children received personalized feedback and advice and were asked to complete a task with questions concerning the advice. Last, teachers were encouraged to organize supportive activities (e.g. games, poster design competitions) Tailoring (TIDieR #9): children completed 2 computer‐tailored interventions, 1 for physical activity and 1 for fat and fruit intake. Both included a questionnaire about demographics, behavior, and psychosocial determinants. After the questionnaire was completed, tailored feedback was displayed on the screen. The first part included general information and normative feedback relating activity and intake levels to recommendations. The next part gave tailored advice based on the theory of planned behavior and the transtheoretical model Modifications (TIDieR #10): not reported How well: planned and actual (TIDieR #11 to 12): self‐administered implementation questionnaires were completed by 1 work group member from each school at the end of the second intervention year. Teachers were asked to rate how well each intervention component was implemented on a scale of 1 to 5. Schools were categorized according to level of implementation: low (n = 3), medium (n = 4), and high (n = 3). BMI Z score increased more in schools with low levels of implementation compared to schools with medium levels of implementation. A significantly greater number of students were boys in the schools with low levels of implementation compared to schools with medium and high levels of implementation Sensitivity analyses: dropout analyses were conducted to compare baseline demographic and behavioral characteristics of students participating and not participating at the 1‐year and 2‐year time points. Data were examined for all participants rather than by study arm. At 1 year, there were no significant differences between participants and dropouts. At 2 years, dropouts were significantly older and consumed more soft drinks than those with data
|
| Outcomes |
All data used in this systematic review were provided by the study authors. The following instruments were used to measure outcomes relevant to this review at baseline, 9 months, and 21 months (end of intervention)
-
Children's dietary intake: 48‐item self‐report questionnaire representing important sources of fat in the Belgian diet, for which the study authors reported reliability and validity; a validated food frequency questionnaire adapted from that used in the Health Behaviour in School‐aged Children study was used to assess fruit intake; a separate food frequency questionnaire was used to assess soft drink and water intake
Available data from the 9‐month and end‐of‐intervention assessments include the following outcomes: total energy intake; per cent energy from fat intake, fruit intake, soft drink intake (used as a proxy for sugar‐sweetened beverage intake), and water intake
-
Children's physical activity levels: total physical activity was measured by summing 3 components from the validated Flemish Physical Activity Questionnaire (a self‐report instrument); other outcomes were measured by accelerometer (instrument not reported) in a subsample of participants
Available data from the 9‐month assessment include the following outcomes: total physical activity, MVPA, light physical activity Available data from the end‐of‐intervention assessment are for total physical activity only; accelerometry data could not be retrieved
-
Children's anthropometry: height measured with a wall‐mounted stadiometer (specific instrument not reported) and weight measured with a Seca scale (Hanover, Maryland, USA)
|
| Identification |
Study name: not reported
Country: Belgium
Setting: schools with technical and vocational education in West Flanders, Belgium
Types of reports: published journal articles
Comments: used the following reports: (1) Haerens 2006b, (2) Haerens 2006c, (3) Haerens 2007a, and (4) Haerens 2007b, as well as unpublished information provided by the corresponding author (Haerens 2019 [pers comm])
Author's name: Leen Haerens
Email: leen.haerens@ugent.be
Conflicts of interest: "the author(s) declare that they have no competing interests" (quote)
Sponsorship source: Policy Research Centre Sport, Physical Activity and Health, funded by the Flemish government |
| Notes |
|
| Risk of bias |
| Bias |
Authors' judgement |
Support for judgement |
| Random sequence generation |
Low risk |
Quote: "...schools were then randomly assigned to the intervention or control conditions..." (Haerens 2007b, p 259)…"
Judgment comment: randomization was computer generated (Haerens 2019 [pers comm]) |
| Allocation concealment |
Unclear risk |
Judgment comment: methods used to conceal the allocation sequence were not described |
| Blinding of participants and personnel
All outcomes |
Unclear risk |
Judgment comment: no information on blinding of participants and personnel was provided. However, given the nature of the intervention, there would have been no way to blind participants and personnel. Also, the children were aged 11 to 14 years, and their performance may or may not have been influenced by lack of blinding |
| Blinding of outcome assessment
All outcomes |
Unclear risk |
Judgment comment: no information on blinding of outcome assessors was provided. Height and weight were measured. Dietary outcomes were measured by self‐report, and physical activity was measured by accelerometer. Body mass index (BMI) and dietary outcomes may or may not have been influenced by lack of blinding |
| Incomplete outcome data
All outcomes |
High risk |
Judgment comment: we calculated attrition rates by dividing the number of children missing data at the assessment time point by the number of children randomized with consent. Rates differed by outcomes and are based on unpublished information provided by the study authors (Haerens 2019 [pers comm]). At the midpoint assessment, total attrition in the relevant study arms was 17.3% (365/2105) for dietary outcomes, 6.6% (138/2105) for self‐reported total physical activity, and 7.2% (151/2105) for BMI. Differential attrition was 13.2% (139/1194 vs 226/911) for dietary outcomes, 2.0% (70/1194 vs 68/911) for self‐reported total physical activity, and 1.0% (78/1194 vs 73/911) for BMI. Data on attrition in the relevant study arms were not available for accelerometer‐assessed outcomes; however, study authors noted that of the 258 children chosen for accelerometry assessment, the parents of 22 children withheld permission and data were missing for 24 children at baseline and for 21 children at the 1‐year assessment. At the end‐of‐intervention assessment, attrition was not available for the relevant study arms for any outcome except BMI. For BMI, total attrition for the relevant study arms was 19.2% (405/2105) and differential attrition was 1.3% (223/1194 vs 182/911)
At each time point and for each outcome, analyses were conducted on matched paired data, reducing the sample sizes. At the end‐of‐intervention assessment, “Drop‐out analyses comparing baseline characteristics of pupils participating and not participating at follow‐up showed few significant differences…Pupils not participating at follow‐up were significantly older and consumed significantly more soft drinks then pupils participating at follow‐up” (quote; Haerens 2006b, p 916) |
| Selective reporting |
High risk |
Judgment comment: the trial was not registered and no protocol was cited that could be retrieved. Dietary, physical activity, and anthropometric outcome data were provided by the study authors in a format that could be entered into meta‐analysis. However, some physical activity outcome data could not be retrieved for the end‐of‐intervention assessment and could not be entered into the meta‐analysis |
| Recruitment bias |
Unclear risk |
Judgment comment: it is unclear whether participants were recruited to the trial after schools (clusters) had been randomized |
| Baseline imbalance |
Unclear risk |
Judgment comment: significant baseline differences were not reported, but the child + caregiver arm appeared to have had more girls and fewer participants from lower socioeconomic status households than the child‐only arm. However, "analyses were adjusted for baseline values, age, and socio‐economic status" (quote; Haerens 2006c, p 850). The study authors did not report on similarities and differences between clusters |
| Loss of clusters |
Low risk |
Judgment comment: no clusters were lost |
| Incorrect analysis |
Low risk |
Quote: "linear mixed models on 1‐ and 2‐year post‐intervention values, with intervention assignment and gender entered as factors, were used to evaluate the intervention effects. School was nested within condition to take into account school variance" (quote; Haerens 2006c, p 850) |
| Comparability with individually randomized trials |
Unclear risk |
Judgment comment: information was insufficient to permit judgment |
| Other sources of bias |
Low risk |
Judgment comment: we detected no other sources of bias |
