Evaluation of a health promotion program in children: Study protocol and design of the cluster-randomized Baden-Württemberg primary school study [DRKS-ID: DRKS00000494]

The program "Komm mit in das gesunde Boot - Grundschule" is a health promotion program for primary school children in Baden-Württemberg, South West Germany, which started in 2009. It is developed, implemented and evaluated by a research team at Ulm University together with a team of experienced teachers. The program supplements the curriculum for primary schools focusing on subjects covering "humans", "nature" and "culture". It is a school-based program to promote healthy lifestyle choices in children, to increase physical, mental, and emotional abilities and, consecutively, to attenuate the increase in body fat and thus to prevent overweight and obesity.

The contents of the program are fully integrated in the school environment: the intervention of the program is implemented in school lessons and during recess (For details see section "Intervention"). A system of particular interested teachers (project delivery consultants) was established to spread the program throughout the state of Baden-Württemberg: it is a federal state of Germany which is situated in South West Germany and has an area of about 35.751 km² with a population of 10.750.000. All project delivery consultants were trained intensively and thoroughly in the program's concept and materials. Based on these central training sessions, these teachers are able to hold local training courses in their region for further primary school teachers. 30 and 34 primary school teachers were trained and certified as project delivery consultants in 2009/2010 and 2010/2011, respectively. These project delivery consultants recruited and trained 453 and 437 primary school teachers in 2009/2010 and 2010/2011 respectively. The project delivery consultants are reimbursed for her/his work; school teachers receive travel subsidies for training lessons and perform the program as part of their routine work.

The advantage of the system is that teachers are trained by a colleague or other teacher and not by an external expert. This enhances acceptance of the program and facilitates translation into the school environment and enables a higher sustainability [36]. Furthermore, such systems provide local training courses which support participation of the primary school teachers because of short distances. Dissemination is facilitated in a whole state approach with limited resources because the scale-cost effects are much lower than in expert intervention programs [37].

The quality of the program is permanently assessed by process evaluation. Additionally, an investigation of effects of the intervention is being performed in the Baden-Württemberg Study. The study protocol was developed in 2009 and approved by the ethics committee of Ulm University in June 2009 (Application No. 126/10). The study started in 2010 with baseline measurements. Follow-up measurements were carried out in 2011. Furthermore, approval of the Baden-Württemberg Study was obtained from the Ministry of Culture and Education of the state of Baden-Württemberg. The teachers and principals of the participating schools as well as the children's parents gave written informed consent to participate. The Baden-Württemberg Study is registered at the German Clinical Trials Register (DRKS), Freiburg University, Germany, under the DRKS-ID: DRKS00000494.

Three parameters are specified as main efficacy outcomes in the study protocol: change in waist circumference, change in subscapular skinfold thickness and change in endurance performance during a 6-min run; changes are measured as difference between follow-up measurement and baseline measurement. These parameters were chosen because of their reasonably objective measurement and the results of the URMEL-ICE study [16]. Changes in a number of parameters concerning the following topics are regarded as further outcomes as well as influencing factors:

The parameters for the further outcomes were chosen because of its importance for the children's health. Details of the parameters are described in the section "Assessments of the Baden-Württemberg Study".

Figure 1 shows an overview and the time course of the Baden-Württemberg Study.

The Baden-Württemberg Study is a prospective, stratified, cluster-randomized, and longitudinal study consisting of an intervention group and a control group. The waiting control group design was realized in this study: in the academic year 2010/2011, the program's intervention was carried out in the intervention group whereas the control group followed the regular curriculum (without intervention). In the academic year 2011/2012, the control group started with the intervention, i.e. after one year time of waiting. Cluster-randomization was performed for participating primary schools in Baden-Württemberg. Blinding is not feasible in such a school based study. Measurements were performed at three different levels, namely level 1 "pupil", level 2 "parents", and level 3 "teacher and principal", see section "Data Collection". Participating children are 1^st grade and 2^nd grade pupils.

Measurements were taken at two points:

Pupils at randomized primary schools whose parents gave written consent for participation in the study were measured at the two points mentioned above.

Additionally, in November 2010 (baseline measurement) and November 2011 (follow-up measurement) a parental questionnaire was issued and returned within six weeks. Teachers and principals were also asked about characteristics of the school environment during follow-up.

A continuation of the Baden-Württemberg Study is planned as an observational study: pupils taking part in the study will be observed until they finish primary school.

At present (winter 2011), all follow-up measurements have been completed. In the next months, data management (scanning of case report form pages and verification of scanned data) and data checks will be performed before starting the statistical analyzes.

The intervention was developed by a team of scientists in collaboration with experienced primary school teachers and is developed for all levels of primary school (i.e. 1^st grade until 4^th grade). It combines elements from behavioral prevention and situational prevention. The three main goals of the intervention are to increase physical activity, to decrease the consumption of sugar-sweetened beverages, and to decrease time spent with screen media. The intervention follows a saluto-genetic, competence, and action oriented approach and is informed by Social Cognitive Theory [38]. The program's intervention is based on one academic year as part of lesson plans and recess. No additional physical education lessons are necessary to carry out the intervention but physical activity is integrated in the whole school day and in the school environment.

The intervention is implemented by the class teacher. For a progressive consolidation of the intervention's main goals from the 1^st grade until 4^th grade, a spiral curriculum is promoted: main topics are recessed from grade to grade. The intervention's rationale and contents are included in folders which contain the manual and materials for the intervention (1^st and 2^nd grade [39], 3^rd and 4^th grade [40]). Each class teacher belonging to the intervention group has taken part in a three-part local training course. The intervention itself includes 20 units per school year regarding the topics "beverages", "physical activities", and "recreational activities". These units are spread over the whole academic year. Furthermore, the intervention consists of two physical activity exercises which are performed every school day ("active breaks", each exercise takes between 5 and 7 minutes). Additionally, "family homework" exercises are given which are small tasks relating to the lesson's topics to involve the parents. Further, samples for parents' evenings and templates for letters to parents in three languages, i.e. German, Turkish and Russian, are included in the folder.

In the academic year 2010/2011, there was no intervention in the control group: although interested, the class teachers belonging to the control group received no local training and no materials for the intervention: they were registered for the participation in the academic year 2011/2012. In the academic year 2011/2012, these class teachers started with the three-part local training course (waiting control group design, see section "Study Design").

Process evaluation is ongoing to assess the quality of central training sessions for project delivery consultants and local training courses for school teachers. To further develop and improve the program, a process evaluation mainly using questionnaires, assessing implementation of and satisfaction with the given materials as well as structure and contents of the training courses is being carried out. Furthermore, information about work load when using the teaching materials and of the basic facilities of participating schools were gathered.

A scientific advisory board was established to support the research team and to help ensure the long-term quality of the program which meets regularly and reviews reports, study plans and results (see appendix).

The collection of data in the Baden-Württemberg Study is realized at three different levels:

In the following section, details of all assessments used in the Baden-Württemberg Study are described. Measurements at level 1 "pupil" are characterized at first.

Measurements regarding the child's level are characterized in the following. All measurements were performed by four regional teams in fall 2010 and fall 2011 as described in section "Organization and Preparation of the Measurements". All measurements regarding the child's level were carried out in the classroom and gym of each primary school participating in the study.

In the following section, details of all measurements on the parental level are described. The variables were assessed using a parental questionnaire.

In the following, details of all measurements regarding the level "teacher and principal" are described. These measurements were carried out at follow-up only.

Two different questionnaires for intervention group and control group were used: in both study groups, characteristics of each school were collected about the number of pupils and teachers, the number of pupils with migration background, the concept of the school (e.g. full-time school, school with mixed-grades), recess format, school building and school environment (e.g. safety of way to school, form of settlement).

Pre-tests were performed to practice the planned procedure of the assessments. At first, in spring 2010 pre-tests for the different measurements at the level "pupil" were performed in several primary schools and kindergartens (final year) to ensure the practicability and feasibility of the assessment tools in children just entered school. For training purposes further preliminary tests in primary schools were carried out by each regional team before starting baseline measurements in fall 2010 and before starting follow-up measurements in fall 2011. Further, three different versions of the parental questionnaire were tested in May 2010 and June 2010 at five primary schools. None of these mentioned schools were later involved in the study.

Data collection at the level of pupils was conducted from the end of the summer vacation until the beginning of the fall vacation in 2010 and in 2011, respectively. For practical reasons four regional teams were established for data collection. Each team consisted of about 10 persons (two scientists and about 8 trained student assistants). The scientists are members of the research group in Ulm, the student assistants were recruited in the respective region. The scientists were trained in measuring anthropometric parameters by external experts. In addition; they were trained in handling and using Actiheart^® activity sensors. Student assistants were trained by the scientists in performing the DKT, interviews, computer-based test with children, and collecting anthropometric data with the exception of skinfolds. Skinfold thickness was only measured by scientists of the research group. Each team was measuring pupils of one or two classes per day; however, each school was measured separately.

The parental questionnaires were issued in November 2010 and November 2011, respectively, and returned within six weeks. Teachers and principals completed questionnaires at the second time in fall 2011.

Information about the program and Baden-Württemberg Study were issued during the academic year 2009/2010 using a number of ways, e.g. education and health authorities, and universities of education; electronic newsletter; television and radio; adverts in training catalogs for primary school teachers; participation at trade shows. The recruitment process was also promoted by ten informative events in different parts of Baden-Württemberg. Further, all primary schools of the state of Baden-Württemberg received written information about the program and the structure of the study, asking teachers to participate. Interested teachers contacted the program center. The participation in the program was voluntary, participating teachers had to agree with randomization. In total, written consent from 172 primary school teachers and principals at 94 schools was obtained (see Figure 2). Teachers who already took part in the program in the academic year 2009/2010 were not included in the study. Because the grade level (1^st grade vs. 2^nd grade) may play a key role for the efficacy assessment, stratification of randomization was carried out for grade level according to the following procedure. Based on information about the distribution of the 172 teachers within the 94 schools, stratification according to number of classes and grade level (grade) was realized for the randomization process: six different school types were identified and used as strata for randomization:

There were 8 teachers at 3 schools who withdrew consent before randomization because of unknown reasons. Thus, randomization was performed for 164 teachers within 91 schools intervention group (IG): n = 45 schools, control group (CG): n = 46 schools). Written consent was withdrawn from 7 teachers within 5 schools immediately after randomization (IG: n = 1 school (reason: too much effort), CG: n = 4 schools (reason: randomization in control group)). Therefore, 86 schools (IG: n = 44 schools (n = 81 classes), CG: n = 42 schools (n = 76 classes)) were included in the Baden-Württemberg Study at baseline. The result of the stratified randomization was the following distribution of classes within both study groups:

Baseline measurements were conducted in 157 classes (i.e. 157 primary school teachers were included). One teacher at a school randomized in a 1^st grade class in the control group withdrew the consent shortly before follow-up (reason: parents' request). However, that school remains in the Baden-Württemberg Study because another class at that school took further part in the measurements at follow-up. Another two teachers randomized in a 1^st grade class in the control and intervention group, respectively withdrew consent at follow-up (reasons: school building burnt down and unknown). Thus, at follow-up the distribution of classes within both study groups was as follows:

In total, follow-up measurements were taken in 154 classes within 84 schools.

In total, 3159 pupils were in the participating classes and 1968 parents (62.3%) of pupils at primary schools participating in the Baden-Württemberg Study gave written informed consent to participate.

Baseline measurements (fall 2010) were taken from 1947 children and data at all three levels ("teacher and principal", "parents", and "pupil") was available from 1670 children (IG: n = 916 (54.9%), boys: n = 452 (49.3%); CG: n = 754 (45.1%), boys: n = 388 (51.5%)). The distribution of these pupils within each grade is shown in Table 1:

The computer-based test for attention and executive function (KiTAP) were performed in a subsample of 513 pupils belonging to 45 classes within 27 schools (IG: 54.8%; boys: 50.3%), objective measurements of physical activity using Actiheart^® were performed in 384 pupils belonging to 56 classes within 32 schools (IG: 55.2%; boys: 50.3%). Results of baseline measurements in both subsamples are presented elsewhere [49, 50].

Figure 2 shows the flow chart of enrollment, baseline measurements, and follow-up in the Baden-Württemberg Study.

Power calculations were made at beginning of the study based on the availability of data for the main outcome parameters at baseline. Three main outcome parameters for the assessment of efficacy are specified in the study protocol: change in waist circumference, change in subscapular skinfold thickness and change in 6-min run performance, changes are calculated as differences between follow-up measurement and baseline measurement. All power calculations are based on simulations of hypothetic data and application of linear mixed effects models [51, 52]. An adjustment according to Bonferroni was made to account for multiple testing [53]. In each mixed effects regression model only group (intervention group vs. control group) was used as explanatory variable and no distinction was made between several classes within the same school. Data from the project URMEL-ICE were used for specifying the differences between intervention group and control group. Randomization data were used for specifying the numbers of schools: IG: n = 44 schools, CG: n = 42 schools. Numbers of baseline measurements available for each primary outcome were used for calculation of pupil numbers in each study group:

Using this information, the following assumptions were made for the number of pupils in each school:

Furthermore, it was assumed that a percentage of 80% measurements available at baseline will be available at the follow-up, too. This percentage is an approximate value and was observed in the project URMEL-ICE.

The simulation results showed that the following values for statistical power could be reached if the assumptions proved right: 85% (change in waist circumference), 76% (change in subscapular skinfold thickness), and 74% (change in 6-min run performance).

Besides descriptive methods, complex statistical models will be used for statistical analysis of main and further outcomes because of the clustered data structure. For example, mixed effects regression models [52] or GEE models [54] will be used. Both procedures can be extended in an easy way to adjust for possible covariates, e.g. variables regarding the familial and social background of children or time span between time of baseline measurement and time of follow-up measurement.

Health-economic aspects will be evaluated in a similar way as performed for the project URMEL-ICE [37].