Multimedia Hydration Education During Physical Education in Schoolchildren

Background and rationale. Adequate hydration supports children's health, cognition, and physical performance, yet many children fail to drink enough fluid, particularly during physical education (PE). A prior six-week trial in the same population, using an adapted board game delivered before PE, improved hydration knowledge and behaviour but allocated only eight classes (clusters), which precluded cluster-adjusted modelling. This trial was designed to address that limitation by using a larger number of clusters and a cluster-adjusted analytical framework, while substituting a scalable multimedia format, deliverable with standard classroom equipment, for the physical board game. The intervention was grounded in the Behaviour Change Wheel and its COM-B model (Capability, Opportunity, Motivation-Behaviour), targeting children's hydration knowledge (capability), the social and physical conditions that normalise drinking during PE (opportunity), and willingness to adopt the behaviour (motivation).

Design. This was a cluster-randomized controlled trial conducted in two public primary schools in Sfax, Tunisia, during the 2021-2022 academic year (February to April 2022). The school class served as the unit of randomization (the cluster) and the individual child as the unit of analysis. All eligible first-grade classes in the two schools were enrolled, giving twelve clusters (six per arm) with an average of approximately seventeen children per cluster.

Participants. First-grade children aged 6-7 years were assessed for eligibility. Eligibility required signed parental consent and the ability to participate in all planned data-collection sessions. Children were excluded for exemption from PE on medical or injury grounds, grade repetition, or absence from one or more data-collection sessions. Of 241 children initially assessed, 204 were retained for analysis (101 in the educational arm across six classes; 103 in the control arm across six classes).

Randomization and blinding. Classes were allocated using a computer-generated sequence stratified by school, with three classes per school assigned to each arm. Allocation at the level of the intact class prevented contamination between conditions. Because the intervention was educational, children and teachers could not be blinded; however, outcome assessors and the data analyst remained blinded to allocation, and questionnaires were scored from anonymised, identification-coded forms.

Intervention. The educational arm received six weekly 15-minute multimedia sessions delivered by the regular PE teacher in Arabic, using slideshow presentations combining short written text, static and annotated images, and brief animations, shown immediately before the physical-activity component of PE. Session content progressed from foundational concepts to applied behaviour and contextual factors. Following the educational component, intervention children completed a standard 60-minute PE session identical in structure and duration to the control arm. Control classes continued usual PE with no hydration content and were offered the programme after the trial for ethical equity.

Outcome measurement. Hydration knowledge was assessed with an Arabic adaptation of a validated 22-item awareness questionnaire, scored as a proportion from 0 to 1, administered immediately before the first session and immediately after the sixth. Water consumption during each session was quantified by weighing each child's individually labelled bottle before and after the session, with a recorded value of zero denoting non-drinking. Acute within-session body-mass change, an indirect indicator of hydration status, was computed from calibrated pre- and post-session weighing. Ambient temperature and relative humidity were recorded at each session as covariates.

Statistical analysis. Analyses were performed in Python (statsmodels). Continuous variables were summarised as median (interquartile range) and categorical variables as counts and percentages. Because intact classes were randomized, confirmatory between-group inferences were derived from models accounting for the non-independence of children within classes, using a random intercept for class and, for repeated-measures outcomes, an additional child-level random effect. Hydration knowledge was analysed by mixed-effects analysis of covariance adjusting for baseline score, age, and sex. Water behaviour was modelled as a two-part outcome (the probability of drinking and the volume consumed among those who drank). Body-mass change was modelled with group, session, and their interaction. Intraclass correlation coefficients were estimated for each outcome, and a cluster-level analysis comparing the twelve class means provided a small-sample sensitivity check. Non-parametric tests were used as further sensitivity analyses.