The prevalence of depression in childhood (10%) indicated relatively high rates in comparison with other samples using DSM-IV depression diagnosis (from 1% to 8%) [51, 52]. However, it is important to note that the prevalence may be overestimated because of the sample characteristics, especially with respect to the nature of the institutions surveyed. Furthermore, these relatively high rates can be explained by the criteria applied for defining the cases of depression and/or by the continuity and variety of risk situations to which most of the children studied are exposed, e.g. poverty, violence, dysfunctional households and difficulty of access to healthcare services. On the other hand, the prevalence verified is equivalent to other studies if only considering clinical cases (6.9%). Using clinical diagnosis, Fleitich-Bilyk & Goodman  found moderate to high overall prevalence of psychiatric disorders in Brazilian community children and adolescents (13%), compared to a British survey (10%); however, no difference was found in relation to depressive disorders.
Furthermore, alarming statistics were revealed for family violence, which could also be explained by the social vulnerability of the families investigated, the cultural acceptance of violence in many Brazilian families and the inefficacy of protection services in the country.
With respect to the scope of the study, namely to examine the association of psychosocial factors with depression in childhood, first of all, there is little evidence of depression according to sex and age in childhood. The sex differences are more related to puberty than to chronological age . With regard to the association with socioeconomic status, studies that enclose populations of different social strata can better explain this issue.
In terms of the individual questions block, the association with externalizing behavior can be explained by (1): comorbidity - as Angold & Costello  emphasize: ‘it is a real characteristic of the phenomenology of child depressive disorders’ (p.155); and (2): exposure to family violence, which is also a consensual risk factor for aggressive and rule-breaking behaviors.
However, according to other works [12, 55–57], all the issues associated with family environment and depression comprise an environment exposed to risk. Parental divorce, a bad relationship between father and child and violence are aspects that are causes of potential depression in childhood. Both factors are interrelated, since there is a tendency for children to remain with the mother after divorce, which may lead to distancing from the father. The feeling of loss, prior or posterior conflicts resulting from separation, fights, and socio-economic aspects are features related to divorce, making the situation even more harmful to the child. The new family organization can facilitate physical and emotional detachment, which reduces family support and induces rejection and hostility.
The finding of a strong link between different types of family violence and child depression, among which psychological violence is highlighted, may indicate that a violent context produces a psychological and emotional imbalance that may trigger the depressive condition. Moreover, it is noteworthy that the low reactive ability of children vis-à-vis depression may contribute to the victimization. Furthermore, the effect of violence can interfere in the prolongation of depression, as the blame, shame, sadness and withdrawal generated by violent situations can all contribute to a depressive constellation that is difficult to revert and foments the condition of victim and depression .
These findings can assist clinical decision-making processes by characterizing psychosocial aspects and guiding educators and families. Efforts should focus on public health models for the prevention of violence and on the development of adaptive coping mechanisms, in accordance with the various stages of risk from the developmental perspective. These focal points should be taken into account in early interventions, especially for those children who come from highly vulnerable families. The effects of violence may alter the timing of typical developmental trajectories. Initially, violence may result in depression and externalizing disorders that cause secondary reactions by disrupting the child’s progression through age-appropriate developmental tasks, and consequently, his/her ability to cope with the social world .
Lastly, future analyses need to focus on investigating the link between depression and the relationship of the child with the mother and siblings, the support from friends and relatives, and sibling violence, since these issues limit statistical results. Besides that, it is important to understand mediators and protective variables in pathways to depression and to determine whether early interventions with children who are victims of violence can reduce the risk of subsequent depression. Conversely, it can be established whether early intervention with children experiencing this disorder can help to reduce the risk of violent victimization. Moreover, children who were exposed to violence, especially those from an underprivileged background, need to be evaluated and treated by trained clinicians. It is also critical to clarify the understanding of the physiological factors, which may indicate the role of genetic and/or early environmental factors in the origins of depression. An approach that takes into account the combination of psychological, family and physiological factors may contribute to the comprehensive course and outcome of depression through interrelationships with the environment.
Limitations of the study
The cross-sectional design limits the findings, which should be considered in the interpretation of the results, since it does not permit to investigate the possibility of reverse causality. The access to only one respondent (the caregiver) introduces a limitation, since only one viewpoint is analyzed. Furthermore, the assessments are retrospective, which can introduce a recall bias. Another limitation refers to the variety of measures, which can generate confounding and interaction, though this was partially minimized by block and univariate analysis. Finally, with respect to the psychosocial factors, the majority of them are not specific to any particular disorder; however the identification of potential factors may indicate aspects that must be considered in the prevention and treatment of mental disorders in children.