Study design and participants

The present study is a cross-sectional analysis of the baseline data from a cluster-randomized controlled trial (NCT03236337) to assess the effectiveness of a physical activity intervention program (MOVI-daFit!) on reducing fat mass and cardiovascular risk and improving physical fitness, EF, and AA among schoolchildren. Ten schools from ten towns in the province of Cuenca, Spain, participated in the study. In all schools, all children belonging to the fourth and fifth grades of primary school (9–11 years old) were invited to participate. The study design, sampling procedures and methods have been fully described elsewhere [22].

The study protocol was approved by The Clinical Research Ethics Committee of the ‘Virgen de la Luz’ Hospital in Cuenca (REG: 2016/PI021). Before data collection, an informational session was conducted in each classroom to explain the study’s objectives, the voluntary nature of participation, the anonymity of responses, and the right to withdraw at any time without consequences. These aspects were communicated in age-appropriate language to ensure the children’s understanding. Following this explanation, each child was explicitly asked to give their verbal consent, in accordance with the procedure approved by the ethics committee. This consent was given in the presence of the classroom teacher, the responsible researchers, and the children’s peers, who served as witnesses. In addition, written informed consent was obtained from the parents or legal guardians of all participants.

For the analysis, only students with complete data—those who completed all assessment tests—were included. Students with disabilities were not excluded solely on the basis of disability; they were eligible if they met the inclusion criteria described in the original trial. Children were excluded if they exhibited difficulties in learning Spanish; if teachers or parents reported serious physical or mental disorders that could hinder participation in program activities; or if pediatricians identified chronic conditions—such as heart disease, diabetes, or asthma—that could prevent participation.

Instruments and study variables

The recruitment period for this study began on September 11, 2017, and ended on October 26, 2017. All measurements were conducted by trained researchers who followed standardized protocols to ensure consistency and accuracy in data collection. The evaluators were blinded to the participants’ group assignments.

Gross motor competence

Gross motor competence refers to the skilled performance of a wide range of motor tasks, as well as the coordination and control of movements involving large muscle groups [23] and is a prerequisite for child physical activity participation and also for engagement in learning and social activities, including sports and games [24]. To assess GMC, two dimensions (aiming-catching and balance skills) of the validated Spanish version of the Movement Assessment Battery for Children-Second Edition (MABC-2) were used [1] for the age ranges 2 (7–10 years) and 3 (11–16 years). This battery has shown strong content and construct validity for assessing MC in children, with test–retest reliability coefficients above 0.80 [1]. The tasks performed were: 1) Aiming-catching 1 (catching): The child had to throw a tennis ball against the wall and catch it with both hands (9–10 years); The child had to throw a tennis ball against the wall and then catch it using only one hand (11 years). 2) Aiming-catching 2 (throwing): The child had to aim a beanbag into a red circle on a mat (9–10 years); The child had to throw a tennis ball into a red circle on the wall (11 years). 3) Balance 1 (static): Single-board balance – The child had to balance on one foot on the balance board (9–10 years); Two-board balance – The child had to balance on the balance board, ensuring that the heel of one foot and the toes of the other foot touch (11 years). 4) Balance 2 (dynamic): Heel to toe – The child had to walk along a line while the heel of one foot touches the toes of the other foot. (9–10 years); Walking backwards heel to toe – The child had to walk backwards along a line, making sure that the toes of one foot touch the heel of the other (11 years). 5) Balance 3 (dynamic): Hopscotch – The child had to jump forward on one leg from mat to mat starting from a standing position (9–10 years); Zigzag hopping – The child had to jump diagonally from one mat to another on one leg (11 years) [25]. Each test received a raw score (number of catches, number of hits on target, seconds on one leg, number of steps on the line and number of right jumps, respectively) where a higher score indicated a better GMC. To enable comparison between tests with different measurement units, raw scores for each test (aiming-catching, static balance, and dynamic balance) were standardized into z-scores. Z-scores were calculated within age- and sex-specific groups by subtracting the group mean from the individual score and dividing by the group standard deviation. The GMC Total Score was obtained as the sum of all standardized test scores.

Fitness variables

Fitness is defined as a set of attributes related to a person’s ability to perform physical activities that require aerobic capacity, endurance, strength, or flexibility and is determined mostly by a combination of regular activity and genetically inherited ability [26]. The evidence-based ALPHA-Fitness test battery [27], designed to assess physical fitness, was used. This battery has demonstrated high criterion validity and reliability across European school populations. Specifically, the 20-meter shuttle run, 4 × 10 meter shuttle run, handgrip strength test, and standing broad jump test have all been validated as reliable field-based assessments of CRF, S/A, and muscular strength, respectively [27]. The following tests were performed:

Cardiorespiratory fitness, using the 20-metre shuttle run test, a valid test to estimate aerobic capacity in children. The children were required to run 20 meters between two lines while keeping pace with the rhythm marked by acoustic signals. The initial speed of 8.5 km/h was increased by 0.5 km/h for each stage (one stage equals one minute). The last half stage completed by the child was considered an indicator of CRF. Maximal oxygen intake (VO₂max) was estimated by applying the Léger formula [28].

Speed-agility, using the 4 × 10 meter shuttle run test. Children were required to run and turn (shuttle) 10 meters twice at maximum speed (4x10m). Two parallel lines were drawn on the floor. When the start was given, the children ran as fast as possible to the other line and returned to the starting line twice, crossing both lines with both feet. Two attempts were made, with an interval of five minutes between attempts, and the best score was recorded in seconds. A lower score indicated a better S/A.

Upper Body strength (UBS), through the handgrip strength test, determined in kilograms using a digital dynamometer with adjustable grip TKK 5401 Grip-DW (Takeya, Tokyo, Japan). The handgrip strength test has high-to-very high construct validity with total muscle strength in healthy children, among other parameters [29]. The test was performed twice with each hand and the arithmetic mean of the four measurements was computed. Handgrip data was normalized by dividing absolute handgrip strength (kg) by body weight (kg).

Lower Body strength (LBS), using the standing broad jump test, which measures explosive LBS. The children jumped horizontally to achieve the maximum distance, and the best of three attempts was recorded in centimeters.

Executive function

Executive function refers to a family of mental processes that have been categorized in three core functions: inhibitory control (the ability to selectively attend, focusing on what we choose and suppressing attention to other stimuli), cognitive flexibility (the ability to change perspectives) and working memory (the ability to hold information in mind and mentally working with it) [30]. Three EFs, using the NIH Toolbox software (NIH Toolbox in Spanish, v. 1.8) [31] were assessed. This tool has demonstrated high internal consistency (Cronbach’s alpha > 0.80), excellent test–retest reliability, and strong convergent validity with standardized neuropsychological measures in pediatric populations [31,32]. All tests were performed individually in a quiet classroom using a tablet (iPad Pro, Apple, Inc., California, USA). The evaluation tests included in the NIH Toolbox for each cognitive domain are listed as follows:

Inhibitory control, using an adaptation of the Flanker test [32]. Participants had to indicate the left-right direction of a centrally shown stimulus while inhibiting their attention to the potentially incongruent stimuli around it. In some trials, the orientation of the flanking stimuli was congruent with the orientation of the central stimulus (>>>>> or <<<<<,) while in other trials, this orientation was incongruent (>> <>> or <<> <<). The NIH Toolbox contained a practice block of trials. If participants passed it, a 20-trial block was presented. These trials consisted of a succession of congruent and incongruent combination of arrows. Using a two-vector method that included both reaction time and accuracy, a final score was calculated for school children who reached a high level of accuracy (>80%), which was as follows: (0.25 x number of correct responses) + 5 – log₁₀ [(congruent reaction time + incongruent reaction time)/ 2]. A total score considering accuracy was determined for children who scored less than 80%.

Cognitive flexibility, using the Dimension Change Card Sort (DCCS) [32]. Participants were shown two target cards and asked to order a group of bivalent test cards, first according to one dimension and then according to the other dimension. After a four-trial practice, children were given a 30-trial block with both ‘shape’ and ‘color’ requirements. Using reaction time and accuracy percentage on preswitch and postswitch, a raw score was determined. Using a two-vector method that integrated both accuracy and reaction time, a final score was calculated for children who reached a high level of accuracy (>80%) as follows: (0.167 x number of correct responses) + 5 – log₁₀ [(congruent reaction time + incongruent reaction time)/ 2]. For participants scoring <80%, a total outcome considering accuracy was determined.

Working memory, using the List Sorting Working Memory Test [33]. Children were given a group of pictures, and items were presented both auditorily and visually. Then, they were required to repeat the names of the items observed in order of size, from smallest to biggest. The number of items shown in each trial increased by one for each series. This test was made up of two parts. In the first one, all the items belonged to the same category (animals or food items). In the second one, the items from both categories (food items and animals) were presented together, and children had to repeat the items by category and size. The List Sorting “Total Score” was composed of final scores based upon a sum of the total correct trials across the two lists.

An EF Total Score, defined as the sum of the z-scores for inhibitory control, cognitive flexibility, and working memory, was calculated following the same standardization procedure described previously for GMC.

Academic achievement

The final grades in language and mathematics, provided by the school administration, were used to determine AA. These subjects were selected because, in Spain, they are considered instrumental subjects, as they serve as a foundation for learning other subjects and help in everyday life and activities. The children’s final grades from the previous academic year were used (range from 1 to 10 score, with 10 being the best grade). Since schools within the same region follow a standardized curriculum, assessment system, and grading criteria, these grades are comparable across students and schools. For the analyses, the arithmetic mean of language and mathematics grades was calculated, representing each student’s academic performance over a complete school year.

Anthropometry

Weight was measured using a scale (Seca 861, Vogel and Halke, Hamburg, Germany) with the child in light clothing and barefoot. For height, a wall stadiometer (Seca 222, Vogel and Halke) was used with children barefoot and standing upright with their sagittal midline in contact with the backboard. Both weight and height were measured twice, and their arithmetic mean was considered for the analysis. Body mass index (BMI) was calculated as weight (kg)/ height (m²).

Sociodemographic data

The age, sex and mother’s education level were gathered using a questionnaire administered to the parents. Data on mother’s education level as the maximum level of education attained by the children’s mothers gathered the validated scale proposed by the Spanish Society of Epidemiology [34] to measure socioeconomic status. This questionnaire includes an item related to mother’s education with six response options: i) no literacy skills; ii) no studies; iii) elementary studies; iv) secondary studies; v) high school; and vi) university studies. In our study, these six categories were collapsed into three, due to the small number of individuals in the lower and upper categories: lower/lower middle (no literacy skills, no studies, and elementary studies), middle (secondary studies) and upper middle/upper (high school and university studies). The mother’s education level was used as covariate in the mediation models (as an interval variable), because it has been shown to be a strong predictor of children’s AA [35].

Statistical analysis

Interval variables were summarized as means and standard deviations. The mother’s education level was expressed as counts and frequencies. Both statistical (Kolmogorov-Smirnov test) and graphical procedures (normal probability plots) were used to evaluate the goodness of fit of variables to a normal distribution and all variables fitted to a normal distribution. Differences in all variables between boys and girls were tested using Student’s t-tests. Effect sizes were estimated using Cohen’s d, with values below 0.20 considered negligible, 0.21–0.49 small, 0.50–0.79 medium, and ≥0.80 large [36].

Pearson correlation coefficients were estimated to examine the relationship between GMC, CRF, S/A, UBS, LBS, EF Total Score, and AA for the total sample and by sex. Values below 0.1 were defined as trivial, weak in the range 0.1–0.29, moderate in the interval 0.3–0.49, and large when greater than 0.5 [36]. A Fisher’s Z-transformation test was performed to assess differences by sex, which were statistically significant [37].

The mediation effect of different fitness components (CRF, S/A, UBS, and LBS) and EF on the relationship between GMC and AA for the total sample and by sex was tested using the PROCESS SPSS Macro, version 4.2, selecting model 6 (serial multiple mediation analysis) and 10.000 bias-corrected bootstrap samples [38]. For this analysis, GMC was entered in the model as the independent variable, the proposed mediators were, consecutively, fitness variables and EF, and AA was included as the dependent variable. Each fitness component was individually entered in a different model: model a, CRF; model b, S/A; model c, UBS; and model d, LBS. In the mediation model, the total (c), and direct effects (a₁, a₂, b₁, b₂, d₁₂ and c´) were estimated. Additionally, this model examines three indirect effects (IEs) (fitness pathway, cognitive pathway, and multiple pathway) that indicate the change in AA for each unit change in GMC that is mediated by each fitness variable and EF. The IEs were considered significant when the 95% confidence interval did not include zero.

The IE is represented by the pathway from GMC to AA via mediators (pathways a, b, and d). The fitness pathway (a₁-b₁), from GMC to AA via fitness variables (a, CRF; b, S/A; c, UBS; and d, LBS). The cognitive pathway (a₂-b₂), from GMC to AA via EF. Finally, the multiple pathway (d₁₂), from GMC to AA via fitness variables (a, CRF; b: S/A; c, UBS; and d, LBS) and EF. Proposed cut points to quantify effect size were 0.14, 0.36, and 0.51 for small, medium and large effect sizes, respectively [39]. The percentages of mediation (P_M) were calculated as (IE/total effect) x 100 to estimate the percentage of the total effect explained by the mediation pathways. All analyses were adjusted by age, BMI, and mother’s education level.

Data analyses were conducted using IBM SPSS Statistics version 28.0 (IBM® SPSS®, Armonk, NY: IBM Corp.), and the level of significance was set at p < 0.05.

Limitations

This study has several limitations. First, its cross-sectional design limits causal inference. Longitudinal studies are needed to confirm the temporal direction of these relationships. Nevertheless, the temporality proposed in our models is consistent with the current knowledge about the relationship between GMC, fitness, EF, and AA [5,10,12,56,57]. Second, while school-based academic assessments are useful, they may not reflect the full range of cognitive abilities or academic skills. Third, we did not account for behavior-related factors such as physical activity or sleep, both of which can influence fitness, cognition, and academic outcomes. Future studies should objectively measure physical activity (e.g., using accelerometers) and sleep quality (e.g., actigraphy or parent-reported sleep diaries) to better understand their potential confounding or mediating roles. Finally, variables such as visual acuity or adiposity were not considered, though both may affect motor test performance [58–60].