The convenience sample consisted of 100 mother-child dyads divided into two Groups: G1 – 50 mother-child dyads composed of mothers diagnosed with recurrent depressive disorder and who had had at least one moderate or severe episode in the two years prior to data collection and no episodes in the last six months and who were receiving outpatient mental health clinic follow-up; G2 – 50 mother-child dyads composed of mothers without psychiatric disorders, who were receiving outpatient Family Health Unit (FHU) follow-up. Mothers aged between 25-45 years with children aged 7-12 years were selected. They were identified through health facilities in the municipal and state healthcare network of a medium-sized city located in an inland city in São Paulo state, Brazil. The children were identified through their mothers.

Inclusion criteria were as follows: diagnosis of recurrent depression for G1 mothers, based on depressive disorder symptoms, according to the criteria of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) (World Health Organization, 2017); absence of mental disorders for G2 mothers, which were systematically evaluated. The exclusion criteria for G1 were mothers who had a single major depressive episode, those who had mild episodes, and those with comorbidities of psychiatric disorders. For both groups, mothers who reported having severe chronic diseases were excluded.

Children of both sexes who scored higher or equal to the low average percentile in the Raven’s Colored Progressive Matrices Test and who lived with their biological mothers were included (only one child per family). Orphans, adopted children, and children with mental retardation and apparent sensory disabilities were excluded to enable control of other variables that could affect their behavior.

To form G1, 7,721 medical records including four years of recorded information were reviewed. A total of 268 mothers diagnosed with recurrent depression with moderate, severe, or remission episodes and who had children between ages of 7 and 12 years were identified. Of the 268 identified mothers, 218 dyads were excluded due to the following reasons: mothers had depressive symptoms at the time data were collected (92), mothers were not located (61), mothers refused to participate (34), mothers had other psychiatric disorders (27), and children had sensory or mental disabilities (4). G2 was composed of 153 women that were identified through a list provided by the FHU and 103 were excluded due to the following reasons: mothers refused to participate (39), mothers had depressive symptoms or psychiatric disorders (34), mothers were not located (27), and children had sensory disabilities (3).

Designed to assess the socio-demographic characteristics of the participants and their families. The socio-economic characteristics were assessed using the Critério de Classificação Econômica Brasil (CCEB, Brazilian Economic Classification Criteria), established by the Associação Brasileira de Empresas de Pesquisas (ABEP, Brazilian Association of Research Enterprises).

Used to evaluate the diagnostic of presence or absence of psychiatric disorders. This semi-structured interview is considered as the gold standard for psychiatric diagnosis. The version used was translated and adapted for the Brazilian population by Del-Ben et al. (2001), who used Kappa Coefficient to test its reliability and found a high value of weighted Kappa (Kw = 0.83). The General Review, SCID I (modules A, B, C, D, E, and F) and SCID II were used for G1 aiming at applying the inclusion criteria established for the mothers. The SCID-NP version of the interview was conducted with G2 mothers for the exclusion of psychiatric disorders. It should be noted that the Diagnostic and Statistical Manual of Mental Disorders – DSM-5 (APA, 2014), currently in its fifth edition, contains the same diagnostic criteria for persistent depressive disorders described in the previous version, which were used for the development of SCID.

Used to measure children’s behavioral problems. It is free to download (http://www.sdqinfo.com) and was developed by Goodman (1997) to screen for behavioral problems in Children and adolescents (4-16 year-olds) in the past six months. It was translated and adapted to the Brazilian population by Fleitlich, Cortázar, and Goodman (2000). It has three different versions (for parents, teachers, and youngsters aged 11 and older); in the present study the version for parents was used. It consists of 25 items divided into four scales that probe difficulties (Emotional Symptoms, Conduct Problems, Hyperactivity-inattention, and Peer Problems and one scale that probes strengths (Pro-social Behavior); these 5 scales have scale five items each. The total difficulties score is obtained by the sum of the scores on the four scales that probe difficulties, ranging from 0-40. The classification as “with behavioral problem” and “without behavioral problems” was based on the cut-off score indicating a problem (total score greater than 16), as proposed by Goodman (1997) for the total difficulties score. The reliability and validity of the instrument in Brazil were evaluated by Woerner et al. (2004), who found good psychometric properties. The test-retest reliability method and Cronbach’s alpha were used to assess reliability, and no statistically significant differences were found between them. The questionnaire was administered at an average interval of approximately 20 days (p = 0.148) with 0.79 of correlation between the scores and intraclass correlation of 0.77 (p < 0.001). To assess validity, the authors compared the Strengths and Difficulties Questionnaire to the Rutter questionnaires and the Child Behavior Checklist (CBCL) (correlations of 0.90 and 0.87, respectively), and also with the diagnostic instrument Development and Well-Being Assessment for Children and Adolescents (DAWBA) (χ² = 13.1, p < 0.001).

Used to identify adverse events and chronic stressors, respectively, in children and families. These scales were proposed by Marturano (1999) and are aimed at detecting adverse events (occasional occurrence in the past 12 months or before) and chronic adversities (since the birth of the child with duration or repeated occurrence for a year or longer) that may have occurred in the child’s life. The Escala de Eventos Adversos (EEA, Adverse Event Scale) consists of 36 items, which can be classified into three domains: (a) school life, (b) family life, and (c) personal life. The test-retest reliability method was used to assess reliability, showing stability indexes of 100%, 97%, and 94%, respectively, between the scale scores obtained in the two times they were applied with a 20-day interval (Santos, 1999). The Escala de Adversidade Crônica (EAC, Chronic Adversity Scale) contains 18 items that encompass chronic adversities related to: (a) child and parents’ health (b) parents’ temperament, and (c) possible family and marital conflicts.

Used to identify resources and possible positive organization patterns in the family environment. It was developed by Marturano (2006) to identify family resources that can be associated with school performance. It has 10 topics related to resources: (a) children’s activities when they are not in school; (b) leisure time shared or promoted by the parents; (c) children’s regular scheduled activities; (d) parent- child shared activities at home; (e) availability of toys and other materials that promote child development; (f) availability of magazines and newspapers; (g) availability of books; (h) parents’ supervision of children’s academic achievement; (i) regular daily schedule; and (j) promotion of family time. In a sample of 100 children having school problems, the psychometric indices (92%-100%) showed good stability in terms of total score, using the test-retest method (Santos, 1999).

Used to identify indicators of positive family organization patterns for the identification of protective mechanisms of child development. It was developed by Pizeta (2014) to evaluate indicators related to the key processes in family resilience, proposed by Walsh (2006, 2016). In the present study, only the interview topics concerning family organization patterns were considered. These topics’ responses were categorized as follows: (1) Flexibility and Reorganization, (2) Routine Stability and Time Management, (3) Parental Involvement in Children’s Daily Routine, and (4) Psychosocial Support and Socio-economic Resources. The reliability of the instrument was assessed using Cronbach’s alpha (0.814).

Performance on this test (at ≥ 25^th percentile) was one of the inclusion criteria for selection of children in this study. It is an instrument used to evaluate children’s level of intellectual functioning, and it was standardized for a representative sample of Brazilian children by Angelini, Alves, Custódio, Duarte, & Duarte, 1999). Construct validity of the test was verified based on the progressive increase in the test results, according to the chronological age of the children evaluated and the internal consistency of the test, which was satisfactory for most items; item-total correlation ranged from 0.30 to 0.80. As for the test precision, the correlation coefficient between the items (split-halves method) for the total sample was 0.92 (Angelini et al., 1999).

After signing the Informed Consent form, the mothers were evaluated by trained psychologists. Data were collected through face-to-face interactions in two or three individual sessions that lasted 60-120 minutes (average of 85 minutes). In the first session, the General Questionnaire, SCID, SDQ, EEA, EAC, and Family Environment Resource Inventory were administered, and in the second session, the Semi-structured Interview was conducted and audio-recorded. When necessary, a third session was held to conclude this interview.

The children were evaluated by trained psychology students in a single session that lasted 10-30 minutes (average of 15 minutes) using the Raven’s Colored Progressive Matrices Test Special Scale.

The data obtained using the SCID and Raven’s Matrices test were used only to select participants to be included in the groups. The SDQ, EEA, EAC, and RAF data were coded according to the technical standards of each instrument. The information obtained by the General Questionnaire was codified to define the children, mothers, and families’ characteristics.

The responses to the Semi-structured Interview selected topics were transcribed fully and literally. These data were categorized and coded following the method proposed by Biasoli-Alves (1998) to interpret the items based on the theoretical framework of developmental psychopathology to identify indicators of positive family organization patterns. These indicators were coded using four previously defined categories with scores ranging from 0 to 42.0 The higher the score, the higher the number of family protective factors. The coding categories were defined by two researchers independently for 30.00% of the transcribed interviews, accounting for 87.27% of agreement.

Data were submitted to statistical analysis using the Statistical Package for Social Sciences (SPSS – Chicago, IL, USA), version 17 for Windows. Initially, the Kolmogorov-Smirnov and Shapiro-Wilk tests of normality were performed. Descriptive statistics was then used. The comparisons between the groups were made using the Pearson’s chi-square test, the Fisher’s exact test (categorical variables), and the Student’s t-test (continuous variables). Univariate logistic regression was carried out to verify the possible influence of the socio-demographic variables that distinguished the groups in terms of the behavioral indicators in the children evaluated. Pearson’s correlation was used to identify possible correlations between the following variables: socio-demographic characteristics of the mother-child dyads, behavioral indicators in the children, adverse events, chronic adversities, family environment resources, and family organization patterns. Spearman correlation was used to identify possible correlations between maternal depression and other variables. The classification system proposed by Maroco (2014) was used to classify the strength of the correlations, as follows: weak (0-0.25), moderate (0.25-0.50), strong (0.50-0.70), or very strong (> 0.70). The significance level of 5% (p ≤ 0.05) was adopted for all tests.