*2.1. Behaviors in Terms of Nutrition and Anthropometric Data*

A reduced and adapted version of the questionnaire used during the 2016 national surveillance study was used in order to evaluate behavior with respect to nutrition habits, as well as to evaluate anthropometric data, in a sample of students attending the third year of primary school (children 7 to 9 years of age) [16,17]. The body mass index (BMI) was calculated and subsequently expressed as a percentile obtained with respect to age-and sexrelated growing reference data provided by WHO for the European region [18]. In order to classify children in terms of weight status categories, the reference percentile ranges provided by the US Centers for Disease Control and Prevention (CDC) were used [19]: underweight (less than the 5th percentile); normal or healthy weight (from the 5th percentile to less than the 85th percentile); overweight (from the 85th to less than the 95th percentile); obese (equal to or greater than the 95th percentile). Additional factors such as general information on physical activity and use of mobile devices were also investigated. The proposed version of the questionnaire was validated during a pilot study in a sample of 100 eight-year-old students (data not reported or included in the study). In the same pilot sample, the reliability index was assessed using Cronbach's alpha (internal consistency coefficient): the alpha value achieved was 0.79, showing a satisfactory level of reliability [20]. Moreover, the modified version of the questionnaire was also validated during the pilot study in terms of intelligibility: the students were asked to assign a score to each item of the questionnaire on a 7-point-scale (from 1: not meaningful to 7: very meaningful). Moreover, in order to guarantee variability in the answers, the original questionnaire was modified in the pilot version: 12 further items (FI) reporting errors (grammatical and/or semantic) were added to the items (OI) belonging to the original questionnaire. OI reported a mean score for each question >6 (almost the maximum); FI showed a mean score ≤1. These data confirmed that the content of the questionnaire was clear to the readers.

The population investigated was represented by eight-year-old children attending the third grade of primary school, selected through a cluster survey design, using the class as a sampling unit, a method also recommended by the WHO and widely used in international surveys [21,22]. First of all, the study was described to the enrolled children with regard to the objectives of the study and the instructions for survey completion. Additionally, parents were fully informed and provided their consent for study enrollment.

The questionnaire was composed of two macro-sections:

(a) Anthropometric form, in which a child's weight and height were reported; measurement was performed at school by LHU healthcare professionals, previously calibrated. Weight was measured to the nearest 0.1 kg by an electronic balance. Children's height was measured to the nearest 0.1 cm by a precision stadiometer;

(b) Children's questionnaire. In the classroom, the children themselves completed the questionnaire, which was divided into three subsections asking:

*First section referring to the day of compilation:*


*Second section referred to the previous afternoon:*


*Third section referring to the previous evening:*


Since the whole reference population included 3620 students, a sample of at least 348 individuals would have been required to investigate the selected variables, assuming a response proportion of 50%, a 95% confidence level, and a 5% margin of error.

The nutritional status in terms of obese/overweight and normal/underweight was compared between the two sex groups using the chi-squared test. A *p* ≤ 0.05 was considered statistically significant.

In order to evaluate the association between variables, a standard binary correlation matrix was used, showing correlation coefficients between variables (a score of 1 is the maximum association coefficient). A table was built in which each cell of the adopted matrix (crossing an abscissa and ordinate value) describes the level of correlation between the variables in abscissa and the variable in ordinate expressed as the Pearson's correlation index, which is the ratio between the covariance and the standard deviations of each pair of variables.
