The aim of the UPRIGHT project is to develop an effective, universal, and holistic school-based intervention. The program is designed to be applied in the school years corresponding to the ages between 12 and 14, regardless of risk condition, and includes their families and the school communities. It is currently being implemented, and its effectiveness tested in different regions from 5 European countries: Basque Country (Spain), Lower Silesia (Poland), Trentino (Italy), Reykjavik area (Iceland), and the regions of North-Sealand, West-Sealand, Funen, North Jutland, and Eastern Jutland (Denmark). The UPRIGHT project is funded by the European Union's Horizon 2020 research and innovation program under grant agreement Nº 754919. The research protocol has been published elsewhere [14].
Here, we present the baseline data collected from all adolescents taking part in the UPRIGHT project before implementing the UPRIGHT intervention. Thirty-four schools from five countries have participated, representing a mixture of urban, rural, socioeconomically disadvantaged, and non-disadvantaged areas.
Data collection
The self-reported information was collected from September 2018 to the end of 2019. All questionnaires had been already validated in all regional languages and were distributed in the classrooms during school hours. They could be completed either on paper or online (Qualtrics, Provo, UT), and a member of the school staff was always present during the completion of the questionnaires.
The data were pseudonymized, i.e., separated from its direct identifiers, so linking the results to a person was only possible using additional information. This information was kept secure and separate from processed data to ensure non-attribution [15].
Measures
The recorded socio-demographic characteristics consisted of gender, year of birth, country of birth, number of children living in the household, and order of birth.
Positive mental health-related outcomes were assessed using the following scales
Mental well-being was measured using the 14-item Warwick-Edinburgh Mental Well-Being Scale (WEMWBS) [16]. Higher scores (score range from 14 to 70) indicate higher levels of positive mental well-being. The levels of mental well-being of students can be interpreted as high (scores > 60), average (scores around 51), and low (scores < 40) [16]. The Cronbach's alpha was 0.85 95% CI [0.84, 0.85] for the current sample.
The health-related quality of life (HRQoL) was assessed employing the 10-item KIDSCREEN scale [17], which is used for children and adolescents aged 8–18. The items assess satisfaction with family life, peers, and school life. The KIDSCREEN-10 instrument provides a singular index for global HRQoL, in the range from 0 to 100. The normative score for the scale is 71.9 [18], and the higher the score, the higher the HRQoL. The Cronbach's alpha reached 0.82, 95% CI [0.82, 0.83], for the current sample.
To measure the principal protective factors of resilience, a 28-item resilience scale for adolescents, READ [19], was used. The READ measures five factors: personal competence (PC), social competence (SC), structured style (SS), social resources (SR), and family cohesion (FC). The total score and each factor score range from 1 to 5. Higher scores reflect higher resilience. Cronbach's alpha of 0.93, 95% CI [0.93, 0.93], was obtained for the full scale.
The 5-item School Resilience Scale [20] measures five interrelated aspects of the school community, considered precursors of resilience and mental well-being of young people: (1) positive relationships, (2) belonging, (3) inclusion, (4) participation, and (5) mental health awareness of all the members of the school community. The total score ranges from 1 to 5, with higher scores indicating higher levels of school resilience. Cronbach's alpha was 0.72, 95% CI [0.71, 0.74].
Mental disorder-related factors were assessed using the following scales
The 4-item Perceived Stress Scale (PSS-4; [21]) are designed to assess the feelings of being overwhelmed and unable to control or predict life events. Scores range from 0 to 16, with higher scores corresponding to higher levels of perceived stress. The norm value for interpreting the PSS-4 scores was 5.43 [22]. Cronbach's alpha reached 0.61, 95% CI [0.59, 0.63].
The 9-item version of the Patient Health Questionnaire (PHQ-9) [23, 24] gives a measure of depression; it also helps assess the severity of depressive disorders. The total score ranges from 0 to 27. The severity of the disorder can be interpreted as (0–4) minimal, (5–9) mild, (10–14) moderate, (15–19) moderately severe, or (20–27) severe. Cronbach’s alpha was 0.79, 95% CI [0.78, 0.8].
The 7-item Generalized Anxiety Disorder Scale (GAD-7) [25, 26] is a screening tool for detecting generalized anxiety disorder. The total score ranges from 0 to 21. The severity of the disorder can be interpreted as minimal (0–4), mild (5–9), moderate (10–14), or severe (15–21). Cronbach’s alpha was 0.84, 95% CI [0.83, 0.84] for the current sample.
The examined conduct problems were the violent behavior and frequency of substance use as measured employing the 8-item screen used in WHO's Health Behavior in School-Aged Children survey (HBSC) [27]. Violent behavior was assessed considering five items measuring the frequency of physical fights in the preceding 12 months, frequency of being bullied or cyber-bullied in the preceding 2 months, and taking part in a bullying or cyber-bullying episode. Substance use assessment included 4 items examining the frequency of lifetime use of tobacco, alcohol, and cannabis.
Statistical analysis
A comparison between female, male, and non-binary genders was performed. Categorical variables were presented as frequencies and percentages (%) and continuous variables as means and standard deviations (SD). The Chi-squared test was used to compare categorical variables and ANOVA for continuous variables. Correlation between the continuous variables was assessed employing the Pearson correlation coefficient and are provided as Online Resource (Table 4). The differences were considered statistically significant at p < 0.050.
A combination of multiple correspondence analysis (MCA) and cluster analysis was employed to characterize the associations between all the mental health-related variables. These two multivariate techniques are widely utilized in medical research to obtain the profile associations based on the similarities of the variables of interest [28].
The MCA is a reduction method characterizing the information for various categorical variables into dimensions explaining the maximum variability levels for the variables included in the analysis [29]. The MCA was employed to identify subjacent relationships between the main variables included in the study. They were included in the analysis as categorical variables, using the categorizations of the scales (explained in the measures section). Continuous variables without any previously defined categorization were grouped into equal segments, obtaining a maximum of five categories. Then, the main dimensions of the MCA were graphically represented in a map, and each category of the variables was plotted as a point. The closer the points are, the stronger the association between the categories.
Cluster analysis was used to divide all participants into groups, based on the main dimensions provided by the MCA, i.e., the association between the variables. The classification was made employing a hierarchical cluster analysis, according to proximity criteria using Euclidean distance, and a k-means non-hierarchical cluster analysis, following the k-means algorithm. The number of clusters was chosen by selecting the best relationship between the appropriate number of clusters observed in the dendrogram and the Calinski–Harabasz index value. To assess the internal cluster quality, cluster stability in the optimal solution was analyzed using Jaccard bootstrap values (100 runs). Clusters were considered highly stable for average Jaccard similarities of 0.85 or higher [30, 31]. The obtained clusters were displayed in the geometrical space constructed using the MCA dimensions. A comparison between clusters was performed; the Chi-squared test was used to compare categorical variables and ANOVA for continuous variables.
Statistical analyses were carried out using the free statistical software R (version 3.6.1); the "ca" package was used for MCA and "stats" and "fpc" for cluster analysis.