The population included in this study comprises primary school students participating in the Chilean Skills for Life I [HpV I] public program. This program carries out promotional and preventive mental health actions for students, their families, and the school community in all the country, while also performing a universal screening of psychosocial risk, including the detection of SM (Vargas & Peña, 2016).

In this national program, questions about psychosocial dysfunction and psychosocial adversity are answered by parents, while the head teacher explains SM. These measures are evaluated in the 1st and 3rd grade, through universal screening.

In 1st and 3rd grade, SM measure allows identifying children who are at risk of presenting SM from those who do not have this difficulty, using cut-off scores and clinical criteria (Guzman et al., 2015; Leiva et al., 2015; Murphy et al., 2015).

For this study, data collection took place between 2013 and 2015. Universal screening included 30,763 children (year 2013), but only those who did not present school difficulties in the 1st grade were considered (N = 26,108), in order to identify variables capable of predicting the emergence of SM two years late (year 2015).

The variables measured in 1st grade allowed identifying those who would participate in the study and knowing if they were exposed to adverse childhood experiences. Then, in 3rd grade it was observed whether the psychosocial adversity measured in 1^st grade predicted or not the appearance of SM in a population that initially was healthy.

In this study, 26,108 parents and 1,927 teachers from 1,418 schools that implemented the HPV program throughout the country evaluated 26,108 first and third-grade students; 50.7% of student were women, and the general average age was 8.39 years in third grade (SD = 0.573).

All participants met the inclusion criteria of not displaying indicators of SM in the first grade and not having participated in a preventive intervention during the second grade.

School maladjustment. The SM was measured using the Teacher Observation of Classroom Adaptation-Re-Revised [TOCA-RR] (George et al., 1995; George et al., 2004), which is the version of the Teacher Observation of Classroom Adaptation-Revised ” (TOCA-R) adapted for Chile (Werthamer-Larsson, Kellam, & Wheeler, 1991).

TOCA-RR questionnaire maintains the psychometric properties of the original instrument. This includes six areas or dimensions through which problems in students’ school adjustment can be observed in the classroom: (1) social contact (Cronbach’s alpha = .886), (2) authority acceptance (Cronbach’s alpha = .910), (3) emotional maturity (Cronbach’s alpha = .762), (4) cognitive achievements for learning learning (Cronbach’s alpha = .913), (5) attention and concentration (Cronbach’s alpha = .912), and (6) activity level (Cronbach’s alpha = .849). This instrument has been used as a criterion for the entrance of children into preventive programs at school and has shown high reliability, having an excellent internal consistency (Racz et al., 2013)

This instrument is completed by the child’s teacher, who rates behavior using a Likert-type scale of observed frequency with scores ranging from 1 = almost never to 6 = almost always. Some examples of questions are: “Hurt others physically,” “Pay attention” or “Have many friends.”

Children who display SM are identified based on their scores on each dimension and following clinical criteria (Guzman et al., 2015; Leiva et al., 2015; Murphy et al., 2015). This measure was applied in 1st and 3rd grade.

Adverse childhood experiences. Eight variables were measured. Four of them are family-related: (1) mother’s adolescent pregnancy, (2) biological father’s absence, (3) mentally ill household member, and (4) insufficient family participation in support networks; two are individual-related: (5) chronic disease that causes children to miss school for long periods and (6) psychosocial dysfunction; and two are school-related: (7) teachers’ negative perception of a child’s behavior and (8) teachers’ negative perception of a child’s academic performance.

The first five variables were obtained from a questionnaire administered to participants’ parents which included five closed questions on the presence/absence of adversities.

Also two school variables were assessed using an appendix of the TOCA-RR questionnaire that explores teachers’ global perception in terms of students’ academic performance and behavior in class through two items (George et al., 2004): (1) “How would you rate the child’s progress as a student?” and (2) “How would you rate a child’s overall behavior in class?”. The teacher responds through a Likert scale ranging from 1 (positive perception) to 6 (negative perception). In this study, ratings of 4, 5, and 6 were interpreted as indicators of a negative perception.

Psychosocial dysfunction. Psychosocial dysfunction was measured using the Pediatric Symptom Checklist (PSC; Jellinek & Murphy, 1988; Jellinek, Murphy, & Burns, 1986). In the United States, the PSC is one of the most widely used and validated screening instruments in different subpopulations.

In the Chilean version (George et al., 1995; George et al., 2004), cognitive, emotional, and behavioral problems of the child observed in his/her home are evaluated. It is made up of 33 questions that are answered by a child’s parents and/or guardians.

Examples of these questions are “He/she complains of pain,” “He/she has little energy”, and “He/she is irritable and angry.” A total score is obtained that determines the presence of psychosocial risk through a cutoff point. Regarding its reliability, the PSC presents a Cronbach’s alpha of .85 for the infant population, as it has a specificity of .68 and a sensitivity of .95 (Jellinek et al., 1988). In the Chilean version, the value of Cronbach’s alpha was .853

In order to test whether the variables studied predicted the appearance of SM in 3rd grade, binary logistic regression, through the odds ratio (OR), was used to examine the intensity of the link between SM and psychosocial factors, while the Wald test was employed to assess the significance of the parameters. The first step was to perform a univariate logistic regression analysis to calculate the unadjusted predictive value of each variable regarding the occurrence of SM in 3rd grade. Afterwards, a multivariate logistic regression analysis was conducted to generate a predictive model of the occurrence of SM using the predictive values of the variables adjusted for the other variables studied. Predictive models were calculated using the enter method selection.

In this study, researchers worked with databases provided by the Skills for Life I [HpV I] national program. The HpV I program informs all participating students, parents, and teachers of the criteria and methodology used. Participation is voluntary. Parents are informed of the administration of the questionnaires and the use of the information collected. They give their consent for their children to participate in the program and for that teachers answers questions about their children.

Students found to display psychosocial risk in the global detection phase are referred to a preventive intervention carried out in the same program or to a specialist if needed. In each of its actions, the program guarantees the anonymity and confidentiality of participant information. With respect to the data used in this study, all personal identification information was removed. Teachers and parents’ responses were matched through an ID number.

This study was approved by the ethics committee of the Faculty of Social Sciences of the University of Chile, institution to which the corresponding author belongs.

Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Three years after the first measurement, it was observed that 12.4% of the 26,108 students who started school without SM displayed maladaptive behaviors in 3rd grade. The percentage of male and female students with such behaviors was similar.

With respect to the frequency of the psychosocial adversity variables, the most commonly observed ones in the total sample were low family participation in support networks (47%) and biological father’s absence from the home (26.6%), while the least frequent ones were teachers’ negative perception of students’ academic and behavioral performance (4.9% and 3.1% respectively).

Comparing the group of students who displayed SM with the group without SM in 3rd grade, the frequency of all variables was observed to increase in the group with SM except for low participation in support networks (Table 1).

Table 1
Frequency of Psychosocial Adversity Variables

All the independent variables other than low family participation in support networks were significant for predicting SM in 3rd grade. Teachers’ negative perception of the student’s academic and behavioral performance variables had the highest predictive values, with odds ratios of 2.07 and 3.68, respectively. The variable with the lowest predictive value was chronic disease of the child (OR = 1.17), followed by mentally ill household member (OR = 1.29). Regarding psychosocial dysfunction, the values obtained show that its presence in 1st grade increased the likelihood of maladaptive behaviors in 3rd grade by 2.42 times, whereas, the absence of the biological father did so by 1.33 times and the mother’s adolescent pregnancy by 1.31 times (Table 2).

Table 2
Univariate Logistic Regression

A general multivariate model was obtained where all the psychosocial variables studied were included (Table 3). In it, all the variables except for chronic disease of the child and low family participation in support networks had significant predictive values. Furthermore, comparing the odds ratios of the univariate analysis with those of the multivariate analysis shows that all predictive values are reduced when adjusted for the other variables; even more so, the variable chronic disease of the child becomes non-significant.

Table 3
Multivariate Logistic Regression. Predictive Model

A new multivariate analysis was performed, but without including the non-significant variables in the model, which yielded the final corrected predictive model (Table 4). The model revealed that, when all these adversity variables appeared together, they increased the likelihood of SM in 3rd grade, with negative perception of behavioral performance and psychosocial dysfunction reaching the highest values (OR = 3.11 and OR = 2.10, respectively).

Table 4
Multivariate Logistic Regression. Corrected Predictive Model

The omnibus test of model coefficients revealed a chi-squared significance below .05, which indicates that the variables included in the model help explain modifications in the dependent variable. With respect to other goodness of fit tests, the log-likelihood obtained surpasses 1, reaching 14,173.640, while Cox-Snell R² and Nagelkerke R² show that the model explains 4.9% and 9.4% of the variation of the independent variable, respectively. The classification table reveals that the model is able to correctly classify 88.1% of the cases analyzed.

Although this effect size is low, for its interpretation the observational design used in this research must be considered that favors ecological validity over internal validity. Thus, although the effect size does not reach the optimal statistical values, in terms of the real world effects that oscillates between 5 and 10% are good indicators of the explanation of social adversities of the appearance of maladaptive school behavior.

This study was carried out considering adverse childhood experiences that prior research has linked to the presence of SM and mental health risks, but which undoubtedly only comprise a fraction of all the psychosocial elements that can act as protective or risk factors during childhood. Therefore, these results only provide a partial view of the phenomenon. Studying SM by incorporating other variables, such as self-esteem, family climate, academic performance, or class attendance, among others, could suggest future research possibilities involving a broader range of protective and risk factors, and may also make it possible to conduct studies with more sophisticated analysis techniques capable of revealing the mechanisms of SM trajectories. For instance, hierarchical or nested linear models could be used in order to estimate the impact of individual risk factors in wider contexts (Dodge & Pettit, 2003).

Similarly, future research could involve more extensive, rigorous, and controlled longitudinal studies capable of examining the trajectory followed by SM and its relation with mental health from childhood to adolescence, but considering the particular characteristics of the psychosocial context of developing countries. In addition, studying the role of teachers and school adjustment/maladjustment in more detail could provide more conclusive information for implementing preventive strategies that assign a key role to teachers as potential receivers and agents of such interventions.

Correspondence: loretoleivab@u.uchile.cl (L. Leiva).