Study design
The Healthy Start-Départ Santé evaluation follows a delayed cluster randomized controlled trial design in which the ECCs are randomly assigned to receive the intervention or serve as usual practice controls. The intervention spans a period of 6–8 months. This period is both preceded and followed by data collections (Fig. 2). Control sites are given the option of receiving the intervention once their participation in the evaluation has been completed. Data collection takes place over three years, such that approximately a third of ECCs recruited will complete the study each year. This study protocol received ethics approval from Health Canada, the University of Saskatchewan, and the Université de Sherbrooke.
Target population and sampling
Participants for the Healthy Start-Départ Santé project are recruited from 61 French and English ECCs in two Canadian provinces: New Brunswick and Saskatchewan. Province-specific recruitment and data collection are carried through a study coordinating center established in each province. Both coordinating centers follow the same protocols. Specifically, a registry of all licenced ECCs in both provinces is obtained. From this list we exclude any ECC which has already received a physical activity or nutrition promoting intervention in the past. This avoids underestimating the effect of the Healthy Start-Départ Santé intervention. To be included in the study, an ECC has to prepare and provide meals for lunch. This is required for assessing the quality of foods being served and for measuring nutritional intake. For feasibility and efficiency reasons, the number of children attending the ECC also serves as an exclusion criterion; centers with less than 20 children between the ages of 3 and 5 are not considered. Information on these criteria is obtained from governmental partners or through a brief telephone administered questionnaire. In order to obtain a final sample of at least 735 children, we estimated that a minimum of 15 children aged 3 to 5 would need to be recruited in 62 ECCs. A minimum of 20 children per center was established to account for a 60 % participation rate (Recruiting 60 % of eligible children in 62 ECCs would give a total of approximately 744 participants). Despite planned measures to minimize losses to follow-up, this issue is inevitable in a longitudinal study of this nature. Based on pilot work, we anticipate an attrition rate of approximately 5 %, which will provide adequate statistical power to conduct the planned analyses. The target of 735 children was based on estimates that 700 (735 - 5 %) participants divided into two groups of 350 participants will provide 80 % power to detect a 10 % between-group difference in outcome considering a within-group standard deviation of 40 %, α = 0.05, intra-class correlation of 0.02, and collinearity between the intervention and other explanatory variables expressed by an estimated multiple correlation of 0.15.
For recruiting purposes, the ECCs are stratified according to province, geographical location (urban and rural) [45] and their respective school district (English or French) (Fig. 3). This stratification process ensures relatively homogenous strata for comparisons, since curriculum can differ between ECCs under different district authorities and that the ECC environment can be influenced by its geographical location [46, 47]. Once stratification is completed, eligible ECCs are randomly selected through a sequence generated using Stata SE statistical software (StataCorp, College Station, TX) according to the above-mentioned strata. Given the large area Saskatchewan represents (652 000 km2), it was decided to carry the study with selected ECCs in the central region in year 1, and in the South and central-North in years 2 and 3. Selected ECC are contacted, provided with information, and invited to participate in the project. Subsequently, ECC directors are telephoned to answer their questions and to confirm their participation while securing the parents’ board support. The recruited ECC are then sent consent forms. Centres which decline participation or with fewer than five children recruited are replaced by other randomly selected ECC from the same stratum. Once the ECC provides final consent, it is randomly allocated to the intervention or usual practice (control) arm. A one to one randomization protocol is applied to each stratum. Invitation packages describing the study and seeking parental consent are sent to parents of all age-eligible children. ECCs return completed parental consent forms to the provincial coordinating centres. Following recruitment of one ECC and its children in the usual practice arm, it was found that it had the same director and shared staff with a nearby ECC that had been recruited in the intervention arm. Given the risk for contamination was regarded as being certain in this situation, the two ECC are considered as one intervention arm ECC, which explains the final sample of 61 ECC enrolled in the study.
Main outcome measures
To assess the objectives specified earlier, each of the measures below is administered to participants in the intervention group before and after the intervention period. The measures are administered at the same times to participants in the control group.
Physical activity level
Children’s physical activity levels are obtained using an Actical accelerometer worn during attendance of ECC for five consecutive weekdays. Accelerometers represent an objective and valid method of measuring physical activity in preschoolers [48–50]. The Actical, an omnidirectional accelerometer, was demonstrated to have higher intra- and inter-instrument reliability than other accelerometers [51]. It correlates at r = 0.89 with directly measured oxygen consumption [50]. For this study, accelerometer data are recorded in 15 s intervals. Data will be cleaned and managed using the procedure recommended by Statistic Canada [52, 53] through a series of publicly available SAS codes adapted for this type of study [54].
Educators’ physical activity levels are measured with pedometers (New Lifestyles SW-200 DIGI-WALKER). The SW-200 pedometers, commonly used in applied research, have consistently shown to be among the most accurate at counting steps in controlled laboratory settings [55, 56], and have demonstrated acceptable reliability in real-life settings [57]. Educators are given a log in which to record the number of steps taken each day during work at the ECC for the same five consecutive weekdays.
Physical literacy
Physical literacy and gross motor skills of children are measured using the Test of Gross Motor Development (TGMD-II). The TGMD-II is a standardized test designed to assess the gross motor functioning of children aged 3 through 10 years [58]. It evaluates two subtests of skills: locomotor (run, hop, gallop, leap, horizontal jump and slide) and object control (ball skills such as striking a stationary ball, stationary dribble, catch, kick, overhand throw, and underhand roll). Of the 12 skills included in the TGMD-II, three (slide, striking with a stationary ball and stationary dribble) are not assessed as they were found particularly difficult and rarely performed adequately by children under six years in pilot work. This was decided in concertation with developers of the TGMD-II. Each child is videotaped while completing the skills. Videos are then reviewed by trained researchers who score two trials for each subtest. Moderate to strong correlations were reported between the TGMD-II and the Comprehensive Scales of Student Abilities, and an inter-rater correlation coefficient of 0.98 indicate good test-retest reliability [58].
Quality of food served to children
Menus and recipes for mixed dishes are collected from each ECC. From these, the types and amounts of meals and snacks served are recorded. Information on food provided is standardized in terms of category and food group as well as number and portion size served.
Food consumed by children
Food intake in ECCs is assessed using the digital photography-weighted plate waste method on two consecutive days. Food consumed is measured by weighing each food served to participants with a digital scale that was manually calibrated (inStyle, 16 × 22 × 1.5 cm, Home Hardware Stores Limited, Canada, 2009) and simultaneously taking a digital picture of the food, and the plate ware on which it was served, with an “ASUS Memo Pad HD7” (ME173X model, Android US, China, 2013). Any leftovers of those foods are then weighed and pictured again at the end of the meal. This is done for every serving. Pictures are taken at a distance of 62 cm and at a 45° angle [59]. This method has been extensively used in studies of school-aged children [60–62] and is considered the most precise measurement of dietary intake [63, 64]. Digital photography has recently been used to characterize meals served and consumed at pre-schools and it is correlated at 0.92 (p < 0.0001) with the weighing method [59]. Information on amount of calories, various food groups, macronutrients, vitamins and minerals consumed is derived by subtracting the weight of leftover of a given food from the weight served. For foods containing multiple food groups, trained research assistants used the pictures to estimate the proportion of food group consumed. This step is assisted by the Food Processor nutritional analysis software (Food Processor, Esha version 10.10.00).
Eating habits and nutritional risks
The Nutrition Screening Tool for Every Preschooler (NutriSTEP), designed and validated in both English and French for 3–5 year-old Canadian children of different ethnicities, is used to assess eating habits and nutrition problems in children [65]. This screening tool consists of 17 questions and takes parents or primary caregivers about five minutes to complete. Assessment of nutritional risk using the NutriSTEP has been validated against registered dietitians’ nutrition assessments (r = 0.49, p = 0.01) [66]. Intraclass correlation of 0.89 (p < 0.001) indicates good test-retest reliability and the majority of items have a Kappa percent of agreement ranging from adequate (κ > 0.5) to excellent (κ > 0.75) [66]. NutriSTEP includes a validated semi-quantitative food frequency questionnaire to assess the usual intake of fast foods as well as the four main food groups according Canada’s Food Guide to Healthy Eating: fruits and vegetables, milk and alternatives, meat and alternatives, grain products. This enables us to estimate the contribution of meals consumed in ECCs to the total intake of the four main food groups.
Early childcare centre environment
Dietary and physical activity practices and policies in ECC are measured using items from the Nutrition and Physical Activity Self-Assessment of Child Care (NAP SACC) [67–69]. The 55-items of the NAP SACC retained for this study collect information on 5 components of nutrition (25 items assessing: feeding environment, feeding practices, menus and variety, education and professional development, policy) and 5 components of physical activity (30 items assessing: time provided, indoor and outdoor play environment, educator practices, education and professional development, policy) [70]. Each item is attributed a score between 0 and 3 points. The NAP SACC is filled out independently by two research assistants in each ECC over the course of 2 consecutive days. Differences in item scores are discussed between research assistants, who then come to a consensus. NAP SACC showed excellent inter-rater reliability with 65 % of items having 100 % agreement and all other items having agreement scores exceeding 71 %.
Knowledge, attitudes, practices and self-efficacy of educators and directors is measured through a self-administered questionnaire. The educator questionnaire collects information on educator practices related to nutrition (12 items, e.g., role modelling healthy food choices) and physical activity (6 items, e.g., incorporating physical activity into classroom routines and transitions) [68], on their self-perceived knowledge of fundamental movement skills (4 items, e.g., level of knowledge of activities that incorporate motor skills in the daily activities of preschoolers), on their self-efficacy (e.g., rate your level of confidence in …) to promote physical activity (1 item), and to teach fundamental movement skills (1 item) [71, 72]. Educators also provide information on their physical activity level through the International Physical Activity Questionnaire-Short form [73], food choices based on an adapted version of the NutriStep, and sociodemographic information (including their age group, level of education, and the length of time that they have been working as an educator).
Complementary measures or co-variables
Beyond the main measures which relate directly to the objectives, the following measures are also administered at the beginning and the end of the study to account for potentially confounding factors in the analyses and to assess secondary outcomes.
Sociodemographic information
Parents provide sociodemographic information through a self-reported questionnaire. Questions in the questionnaire are drawn from Statistic Canada’s Canadian Community Health Survey [74]. The questionnaire provides information on parents’ education levels (highest level of education completed by each parent), family structure (number of siblings and family status) and socioeconomic level (estimate of total family revenues, before deductions, in the last 12 months). Parents also report their physical activity levels and food choices based on the same questions as the educator questionnaire. Parents’ perception of their child’s physical activity abilities are assessed through 4 items of the Perceptions of Physical Activity Importance and their Children’s Ability Questionnaire (PPPAICAQ) [75]. The NutriStep questionnaire is also completed by parents.
Body composition
Height, weight and waist circumference of children are measured using a standardized protocol [76]. Two measures of height (SECA Stadiometer – Model 213) to the nearest 0.1 cm, weight (SECA Scale – Model 761) to the nearest 0.2 kg, and waist circumference to the nearest 0.1 cm are obtained for each participant. If discrepancies greater than 0.5 cm for height and waist circumference or 0.2 kg for weight are observed between the two measures, a third measure is obtained. The average of the two closest measures is recorded. Body mass index (BMI), calculated using the ratio of weight (kg) and squared height (m2) will be used to determine if children are overweight or obese by following the International Obesity Task Force age-adjusted thresholds as recommended [77].
Training of research assistants
Before every data collection periods, research assistants are given a full day of training on data collection and entry to ensure that all procedures are standardized across sites. Research assistants are trained specifically on how to collect anthropometric data, how to conduct plate waste analysis, and how to complete the environmental scan. They also practice all fundamental movement patterns so that they can demonstrate them correctly, and review how to properly videotape children’s movements. Databases are also presented, and research assistants are trained on how to enter all data collected. Provincial study coordinators oversee all aspects of data collection and data entry to ensure protocols are applied as intended.
Data analysis
As a first step, each variable distribution will be examined in order to manage obvious errors such as outliers or questionable input. If necessary, appropriate transformations will be applied. If there are systematic missing variables, these will be filled using multiple imputation methods. After the initial data cleaning and processing, descriptive analyses will be used to characterise the two groups of participants and their context.
Analyses of the main study objectives will adhere to the intention-to-treat principle. Multiple linear regression will be used to determine if, in comparison to participants in the usual care group, participants exposed to Healthy Start-Départ Santé had greater increases in ECC-provided opportunities for physical activity, ECC-provided opportunities for healthy eating, in physical activity levels, in healthy eating behaviours, and if they had greater improvements in physical literacy. A multilevel approach will be used in each model to account for clustering that could result from children attending the same ECC. Potentially confounding variables (i.e., age, parental education, household income, etc.) will be included in the models.
Limitations
Educator turnover is common in ECC, meaning that new educators in centers offering the Healthy Start-Départ Santé intervention may not receive the intended training. Two elements of the intervention nevertheless reduce this possibility. First, one booster session per centre is offered as part of the intervention to reinforce the training of educators who are there since the beginning and to ensure that others are trained to implement Healthy Start-Départ Santé. Second, Healthy Start-Départ Santé is designed to be integrated into centres’ everyday functioning so that all employees should be exposed to it naturally. Also, children’s physical activity level and dietary intake can vary from one day to the next. This could cause the data to be non-representative of typical behavior. Data collection over five consecutive days for physical activity and two consecutive days for dietary behavior will help minimise this risk.
Trial status
Participant recruitment began in autumn 2013. Intervention delivery and data collection began at the same time and are ongoing. This trial is registered through the ClinicalTrials.gov registry (Clinical Trials ID = NCT02375490; Protocol ID = 6282-15-2010/3381056-RSFS).