Violência familiar na infância, violência no namoro e saúde mental na adolescência

Abstract

The main objective of this thesis was to know the magnitudes and patterns of dating violence in adolescence (DV) and family violence against children (VAC), the repercussions of on different DV patterns, and the relationship of VAC patterns to health mentality in adolescence. The general objective raised three specific goals. Each one was explored in one manuscript. The data used in the thesis were derived from the research “Rape of the vulnerable and other violence against female adolescents and young people”, developed in 2017 and 2018, with 721 students from public and private schools in the IX Administrative Region (AR) of Rio de Janeiro. The Brazilian versions of the Conflict in Adolescent Dating Relationships Inventory (CADRI) instrument, the Childhood Trauma Questionnaire (CTQ) and the General Health Questionnaire (GHQ-12) were used to identify DV, VAC and to estimate mental health trends respectively. Data analysis strategies varied according to different objectives/manuscripts. Manuscript 1 estimated the magnitudes of exposure and perpetration of DV among students from the IX AR in Rio de Janeiro. The estimated victimization prevalence ranged from 16.7% (sexual) to 94.6% (emotional). The prevalence of perpetration ranged from 9.9% (sexual) to 94.6% (emotional). Adolescents who reported child abuse, those living in violent areas, and those who consumed alcohol more frequently had a higher prevalence of DV. Manuscripts 2 and 3 used latent classes analysis (LCA) to identify the DV patterns (manuscript 2) and the VAC pattern (manuscript 3) based on the aggregation of individuals into classes, characterized according to the probability of endorsement to the different items from the CADRI and CTQ instruments. In addition to identifying these patterns, we sought to use the estimated classes in causal models, assessing the effects of the VAC dimensions for the DV patterns (manuscript 2) and the VAC classes for adolescents' mental health (manuscript 3). In manuscript 2, three latent classes of DV were identified among girls and boys. Childhood sexual abuse increased the chance of belonging to the most severe class of DV victimization among girls (OR = 5.72; 95% CI = 2.51-13.06). Emotional (OR=5.97; 95% CI = 1.95-18.30) and physical (OR=3.16; 95% CI=1.07-9.34) abuse increased the chances of classifying boys in the most severe class of DV perpetration. The LCA of manuscript 3 also identified three classes of VAC, both in girls and boys. Despite several similarities between the classes of the two sexes, girls suffered more with sexual abuse and boys with corporal punishment (physical punishment). Belonging to the most severe class of VAC increased the GHQ-12 scores by 2 to 3 points compared to the milder classes of violence. Based on these results and with the discussions brought by the three manuscripts, it is expected that this thesis has contributed both to the thematic discussions of violence and presented some methodological advances in the field of epidemiological analysis. Regarding the first aspect, once again, it is evident that violence between parents and children is a public health problem due to its high frequency and direct and specific relationship with violence in adolescents' dating. Debates on the relationship between violence and gender have expanded, as both the prevalence and patterns of violence differ between boys and girls since childhood; patterns of violence were identified that indicate that experiencing different types of abuse is the most frequent situation among children and adolescents, and it was noticed that the repercussions of violence against children on the mental health of adolescents depend on the pattern of co-occurrence of different forms of child victimization. Regarding methodological advances, the thesis is innovative as it proposes the use of LCA to characterize VAC and DV for the first time in Brazil. Furthermore, the thesis advances by articulating the identification of patterns of violence via the analysis of latent classes with their risk factors and consequences, expanding the possibilities of inference from the models.