Locally implemented prevention programs may reverse weight trajectories in half of children with overweight/obesity amid low child-staff ratios: Results from a four-year prospective study in France

Background The aim of the present study was to prospectively assess 4-year changes in the weight status of children between school-based prevention programs locally implemented in French municipalities with the same organizational support, and focused on experiential learning interventions promoting healthy eating and physical activity. Methods Using a body mass index chart established by the International Obesity Task Force, school nurses assessed first-grade children for weight status (normal, overweight, obesity) several weeks before the launch of each program in 2011, and 4 years later, together with the BMI z-score measuring overweight/obesity severity. Child-staff ratios (CSRs) by occupation (canteen seervice/extracurriculars) and training session (healthy eating/physical activity) were computed in each municipality. Results During the 4-year follow-up period, weight status improved in half of the children with overweight/obesity, and worsened in 6.6% of children with over/normal weight. In children who remained overweight, the BMI z-score diminished over time. Estimates of the positive 4-year weight change increased with age and were significantly higher in low-to-moderate CSR multicomponent programs ( City3 and City4 ) compared to moderate CSR single-intervention programs (reference: City1 ). The high-to-moderate CSR multicomponent program ( City2 ) had a similar effect as the reference. The estimated negative weight change decreased with age. Conclusions Training ancillary school staff to experiential-focused interventions for healthy eating and physical activity in locally implemented school-based programs contributed positively to preventing childhood obesity without interfering with educational activities. However, low CSRs seem to be pivotal for optimal outcomes, especially in schools in deprived areas.

3 Low child-staff ratios seem to be pivotal for optimal outcomes, especially in schools in deprived areas.

Background
Obesity in childhood is difficult to reverse(1) and often persists into adulthood, causing many health problems (2). According to systematic reviews, school-based prevention programs may be effective in promoting healthy behavior (3,4). They typically include educational, environmental, and social activities designed to improve dietary habits and reduce sedentary time in schoolchildren (5,6).
However, even sound evidenced-based interventions yield modest effects on behavior and adiposity measures (7,8).
Insights from implementation studies may partly explain these mitigated outcomes. First, how authorities bring prevention programs into the school communities may dramatically influence local dynamics. In France, the Education, Health, and Territory (EST) program included the core principles of health-promoting schools, such as training and support of staff to develop school health policy, focus on school environment and adaptation to local context, community involvement, and development of health-related knowledge, skills, and competencies,(9) but decisions and approvals came from upper authorities, and this "top-down" approach created a reluctance to participate locally. (10) In addition, health education programs require teachers to acquire additional competencies (11) and to include health education in their curriculum. (12) However, teacher work is already described as increasing in complexity and intensity because of societal changes, reformed and increased work tasks, and multitasking. (13) Urgent unforeseen priorities, competing responsibilities, and high workload may also constitute barriers to successful implementation, (14) especially when training and activities seem complex or theoretical. (10) The success of school-based prevention initiatives involves balancing evidence-based interventions with the flexibility to permit local educational communities to target their specific needs. (15) Fostering commitment entails giving local stakeholders the freedom to shape their own programs and providing training and materials to persons interacting with children, (16) which may include both teachers and ancillary staff.(10) Following these core principles, some associations provide organizational backbone support (17) to local communities in charge of the education and care of children. This alternative approach to downstream interventions may induce variation between locally implemented programs, warranting a thorough examination of the relationship between process and outcome indicators. (18) The aim of the present study was to prospectively assess 4-year changes in the weight status of children between school-based prevention programs locally implemented with the same organizational support and focused on experiential learning interventions promoting healthy eating (HE) and physical activity (PA). Another objective was to investigate process indicators of each locally implemented program and assess the influence of process indicators, expressed as child-staff ratios by occupation/component, on these changes.

Intervention
The Vivons en Forme (VIF; "live healthy") organization is a continuation of the obesity prevention scheme previously known as Epode, (19) a community-based prevention program aimed at promoting healthier lifestyles among children and their families, and involving municipal services in charge of child education and care under the supervision of a local coordinator. However, compared to Epode, the non-governmental organization acting as a backbone structure changed its process in 2010, following four new pathways in order to improve program efficiency. First, the name of the program was changed in order to be better accepted by the local stakeholders, including families and children, removing the mention of obesity in the name of the interventions Second, a full social marketing approach was included for each yearly implemented thematic. (20) Third, toolkit materials were pilottested in living labs to collect input from users and stakeholders before application in real-life settings, and in the participating cities. (21) Lastly, the implementation process was centered around local stakeholders, including but not limited to school staff, as well as participation and empowerment. (22) The principle aim was to foster self-efficacy and a long-lasting effect in local school staff newly involved in the field of prevention and health promotion. Local stakeholders have the freedom to shape their programs, and can request additional interventions during the course of the program. The basic underlying principle of this "choose-and-pick" approach was to foster staff involvement and sustainably change their interactions with children and parents. Each participating municipality applies for a minimal 5-year period, and their representatives have to regularly attend regional coordination meetings to receive up-to-date information on training sessions and tool upgrades.

Study design and participant selection
To meet the study objectives, a prospective design was used. Only the four municipalities that systematically monitored schoolchildren's weight status were invited to participate. In these municipalities, VIF counselors (an engineer in nutrition and public health, a sociologist, and the leading coordinator of the program organized training sessions for the municipal staff in charge of canteen service and extracurricular activities (ECAs) in primary schools. Training sessions and toolkits integrated roadmaps for conducting interactive activities with the children and reinforce child-staff interactions via concrete experiences (Table 1). Brochures highlighting the beneficial effect of HE and PA for children were systematically provided to parents. (22) They included tips on how to help kids stay hydrated by drinking water, on breakfast preparation, food breaks (including fruits), avoiding snacking between meals, on treats and smart portion sizes, and how to easily cook healthy meals at low cost.

School nurses used a body mass index (BMI) chart established by the International Obesity Task
Force, which allows classification of children into weight categories(23) (i.e., underweight, normal weight, overweight, and obesity), to assess the weight status of first-grade children at the school premises several weeks before the launch of each program in 2011. Children wore light clothes and no shoes during the weighing sessions. In addition, BMI Z-scores were computed using BMI-for-age reference standards (24) in order to account for overweight/obesity severity. Baseline and follow-up weight status, with BMI z-scores, were matched for gender and age at inclusion, and whether children were schooled in a zone of priority education (zone d'education prioritaire, ZEP) was indicated. ZEP refers to schools in deprived, usually urban, settings that are earmarked for special state support. The decision to categorize a school as a ZEP lies with the administrative authorities, who can release additional funding to finance special needs education. A follow-up weight assessment using the same methodology was performed among the same children in 2015. When it comes to the process evaluation, the number and occupation (canteen service or ECAs) of persons who attended training sessions between 2011 and 2015 were systematically recorded by thematic component (HE and/or PA).

Data blinding and confidentiality
A study number was attributed to each municipality (City#) and each child in the database to ensure confidentiality. The final database was completed in 2016, but anonymized data were transmitted to researchers in charge of statistical analyses in 2018 due to the administrative authorization procedure in each participating city.

Statistical analysis
Process indicators were expressed as numberand occupation of school staff attending training sessions by thematic component in each municipality, and then converted into child-staff ratios (CSRs), the number of children for each trained staff member, by occupation and thematic session.
Because an average ratio of 8 children per adult was found in early childhood education and care settings,(25) the CSR was classified as "low" if between 1 to 5 children per adult, "moderate" if 6-9 children per adult, and "high" if > 10 children per adult. Categorical data were expressed as numbers and percentages and compared using the chi-squared test. Numerical data were expressed as means and standard deviations (SDs) and compared by one-way analysis of variance or the non-parametric Wilcoxon comparison test. Outcome indicators were 4-year changes in weight status, which were considered "positive" if obesity changed to over/normal weight or if overweight changed to normalweight, and "negative" if normal weight changed to overweight/obesity or if overweight changed to obesity. To investigate the influence of process indicators on weight changes, we entered the CSRs (low, moderate, high) by occupation (canteen service/ECA) and thematic component (HE/PA) as interaction terms in a logistic regression using positive 4-year weight change as the binary outcome (yes/no), with adjustments for age at inclusion, gender, and school area (deprived/non-deprived). The same statistical procedure was employed with negative 4-year weight change as a binary outcome in children characterized as over/normal weight at inclusion. Estimates were expressed as odd ratios (ORs) with 95% confidence intervals (CIs). Statistical analyses were performed using the SPSS statistical package, version 20 (SPSS, Chicago, Illinois, United States).

Results
Of the 850 first-grade children schooled in these four municipalities, 23 (2.7%) were not enrolled or removed from the database based on parental request. In 2011, the 827 children included in the analyses ( Table 2) were gender-balanced, with an average age of 6.4 years (SD = 0.79) and 59.3% schooled in deprived areas. The mean age was significantly higher in City1 compared to the other municipalities (p < 0.001), whereas children schooled in deprived areas were significantly overrepresented in City4 (77.4%) and City1 (71.6%) compared to City3 (45.3%) and City2 (0%; p < 0.001). The distributions of gender and weight status were similar between municipalities. Over the 4-year study period, City1 requested two training sessions on HE for canteen service staff, whereas City2, City3, and City4 requested 3, 5, and 15 training sessions, respectively, on HE and PA, for staff in charge of the canteen, and those in charge of ECAs. CSRs are given in Table 3.
When CSRs were entered as interaction terms in multivariate models, they yielded comparisons between the four participating municipalities, with City1 as the reference (moderate CSR single-component intervention). In the 169 children with overweight/obesity at inclusion (Table 5), the estimated positive 4-year weight change increased with age and was significantly higher in low-to-moderate CSR multicomponent interventions (City3 and City4) compared to the reference. The high-to-moderate CSR multicomponent program in City2 had a similar effect as the reference program in City1. In children with over/normal weight at inclusion (N = 795), the estimated 4-year negative weight change decreased with age and was unrelated to the process indicators being studied.

Discussion
During the 4-year follow-up period, weight status improved in half of the children characterized as overweight/obesity. In children who remained overweight, the BMI z-score diminished significantly over time, and being schooled in a deprived area had a negative, though not significant, influence.
The implementation mode under study was similar to traditional school-based prevention programs in that it provided training and materials to local stakeholders.(9) However, the VIF program provides tools previously City3 and City4, compared to 4 in 10 in City 1 and City 2, and approximatively 38% over a 9-year period at the national level. (28,33) Many studies have investigated the relationship between CSRs and outcomes in childhood education and care, mostly on child cognitive and emotional development, but the first attempt to systematically review and meta-analyze this highly complex and heterogeneous literature revealed few, if any, relationships. (18) These process indicators are often overlooked in obesity prevention program evaluations, (7) and the reasons behind their variations at the local level warrant further examination. However, they seem pivotal in childhood obesity prevention programs, though their optimal values still remain to be determined. Finally, it seems that older age had a sound positive influence on 4-year weight changes, though the children were relatively close to one another in regards to age (mean 6.38 years, SD = 0.76). Minor differences in this life period could mark the transition between two milestones of cognitive development, but the complexity of developmental theories warrants caution. (34) If confirmed elsewhere, this result would advocate, at the very least, for including 2nd grade children in these programs.
The present study has limitations related, in part, to the implementation mode under examination. Avoiding prescriptive approaches may have contributed to fostering local dynamics, but also precluded comparisons between balanced interventions. Furthermore, interactions between parents, children, and municipal school staff in charge of PA and HE were targeted by the programs but not directly assessed in the study. In addition, the municipalities participating in the study may have differed from other cities in France, as they systemically monitor the weight status of schoolchildren under their supervision. Considering the present findings, low CSRs in childhood prevention programs could be even more important in deprived areas. (35) This proportionate universalism(36) warrants further examination in relation to parental involvement and other variables of interest.
(31) Nevertheless, this prospective study has some methodological assets, as it relies on reliable estimates of weight status collected twice at a 4-year interval among the same children at primary schools from distant municipalities and compared, which avoided contamination.

Conclusions
Training ancillary school staff to experiential-focused interventions in locally implemented school programs contributed positively to childhood obesity prevention without interfering in educational activities. However, low CSRs are pivotal for optimal outcomes, especially in deprived areas, and the reasons for variations warrant further investigation.

Declarations
Ethics approval and consent to participate The study protocol was approved by the CNIL (Commission Nationale Informatique et Libertés), a governmental body in charge of checking compliance with ethical and individual data protection regulations in France. The CNIL fulfils the role of the Institutional Review Board in France for non-invasive research. Children's weight status and other variables were available from school medical service records, and analysed after anonymization without Institutional Review Board oversight, as legally permitted in France (Article L. 541-1 du code de l'éducation) and in the United States (https://www.hhs.gov/ohrp/regulations-and-policy/decision-charts/index.html). The study's protocol was approved by local authorities in charge of children's education and care (municipal authorities, school directors, and school health services). Verbal consent was obtained from the parents, prior to commencing the study. They received written information on the study objectives, about their right to accept or refuse research participation, and how to withdraw the child from the study once it has started, if they wished to do so.
School nurses asked for the child's permission before assessing weight status.

Consent for publication
Not applicable

Availability of data and materials
The dataset generated and analyzed during the current study is available in the Open Science Frame Network   Note: NS = non-significant; SD = standard deviation

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download. STROBE_checklist_cohort.docx