The study was carried out in the city of São Gonçalo, located in the metropolitan region of the state of Rio de Janeiro, Brazil. The city has a population of over 1 million people, of which one-third are children and adolescents. The city is characterized as a low resource area with low indicators of sewage and urbanization, low per capita income (20% of the population earned $79.4 or less in 2016), lack of attendance to school (12.2% of 15–17 years old adolescents out of school), high rates of urban violence (e.g., the homicide rate is 43.9 per 100 000 inhabitants), and child mortality rate was 12.75 per 1000 live births in 2017 [17–18].
Multi-stage cluster sampling was performed based on a population of 3487 teenagers in 9th grade classes (the last year of fundamental school before entering high school) in public and private schools in 2010. Individuals whose ages were not possible to be accessed or were younger than 13 and above 19 years old were excluded. Sampling followed two stages: (1) schools, with a probable proportional number of 9th grade students, and (2) classes, with the application of the questionnaire to all students present. The sample was designed to obtain proportion estimates with a 95% Confidence Interval (CI) level, an absolute error of 1.6%, and PTSD prevalence in 8% . The estimated sample was 1105 students, and 1129 participated in the study.
A total of 43 public and 30 private schools participated in the study, generally having two classes per school. After the school's approval, the parents signed an informed consent and the adolescents an informed assent. The procedure was explained to the adolescents by experienced researchers and self-administered questionnaires were applied in the classroom. A mental health professional was available to support the students if necessary. The questionnaire was tested in four public schools (51 questionnaires collected) and three private schools (46 adolescents). Students in these schools answered the survey twice whithin seven days, to test its reliability.
Data analysis showed that 918 students experienced at least one traumatic event, essential condition for PTSD according to DSM-IV. In total 862 (65% female and 35% male; Mage = 15 years old) had complete information about PTSD, and that is the number of adolescents analyzed in this study. The sample representation was examined by comparing the social stratification in the analyzed sample (n = 862) to the initial one obtained (n = 1129). No statistical difference was found between the groups, p = 0.238).
The self-administered structure questionnaire was applied in the classroom with an average duration of 60 minutes. The questionnaire was composed by validated scales previously used in international studies with Brazilian population:
Sociodemographic profile. Sex, age, skin color, religion, parental education, and income. The income aggregates social stratification groups scored as upper/middle and lower social strata [20–21].
Family Violence. The Conflict Tactics Scale [21–23] was applied to measure severe physical violence committed by mother and or father against the adolescent, such as kicking, biting, hitting, spanking, burning, strangling, suffocating, threatening with a knife or a gun. At least one positive answer indicates severe physical violence. Satisfactory internal consistency was found for physical violence by the father (Cronbach's α = 0.69) and by the mother (Cronbach's α = 0.73) against the child.
Psychological violence was investigated using the Empirically scaled measures of psychological/verbal control and physical/sexual abuse of Pitzner e Drummond [24–26], characterized by 18 items that reflect acts committed by significant people (family included) against the child studied, such as humiliation, criticism, and use of abusive names such as "crazy," "idiot," or "stupid". A deviation above the mean was the cut-off point adopted to define adolescents victimized by psychological violence. Excellent internal consistency was showed (Cronbach's α = 0.93).
Sexual violence was assessed through the question "Did your relationship with your parents involve any kind of sexual experience?". Answer options were ‘yes’ or ‘no’. This item was developed for this study.
School and Community Violence was measured using the Self-Reported Offenses that evaluate the exposure to violence at school and in the community in the past year [27–28]. The eight items with ‘yes’ or ‘no’ answers include humiliation, threatening, aggression, damage to personal belongings, contact with firearms or knives/cutting instruments, having been robbed, and having had money taken by force in both contexts. The answers were categorized according to absence, presence of one event, or presence of two or more events. The Kuder-Richardson's coefficient was 0.52 to school violence and 0.57 to community violence, which is acceptable when we consider the low numbers of items in each scale .
Social support. The scale Social Support Battery developed by Sherbourne and Stewart [30–32] was used. It consists of 19 items and five dimensions: a) emotional (support received through trust, being listened to, sharing concerns/fears, and understanding problems), b) information (receiving suggestions, good advice, information, and desired advice), c) material (someone available to help in case the adolescent has to stay in bed, be taken to the doctor, prepare meals, and help with daily tasks in case of sickness), d) affective (someone demonstrates affection and love, gives a hug and loving), and e) positive interaction (someone to have fun together, relax, do nice things and distract the mind). Good internal consistency was showed for affective support (Cronbach's α = 0.82), positive interaction (Cronbach's α = 0.85), emotional support (Cronbach's α = 0.85), and information (Cronbach's α = 0.84). Due the little number of items only material subscale showed lower internal consistency (Cronbach's α = 0.57). Kappa ranged from 0.099 to 0.744. The scale was used continuously.
Social Functional Impairment. Brief Impairment Scale (BIS) [33–34] was used to evaluate the overall child and adolescent impairment, based on parents' vision. It consists of 23 items that include three areas: interpersonal relationship, school functioning, self-care, and self-fulfillment. Scores above 15.5 on the full scale are considered positive for overall functional impairment. Originally, the total scale's internal consistency varied between 0.81 and 0.88, and between 0.56 and 0.81 in the three subscales . Previous studies reported high convergent and concurrent validity for BIS . Adequate internal consistency was found (Cronbach's α = 0.72)
Resilience. The Resilience Scale developed by Wagnild and Young scale [35–37] was used to evaluate positive psychosocial adaptation levels to adverse life events. It has 25 items, and good psychometric indexes have been obtained for a teenage sample. Adolescents whose items sum was below one deviation from the mean are considered not resilient (73.0 ± 13.0). Adequate internal consistency was found (Cronbach's α = 0.78).
Post-Traumatic Stress Disorder diagnostic. The UCLA - University of California at Los Angeles Post-Traumatic Stress Disorder Reaction Index for DSM-IV for adolescents was used [38–40]. The instrument is divided into three parts: (1) 14 items to assess traumatic life events. The item regarding earthquake was replaced by landslide to adapt to the Brazilian context; (2) 13 items to assess subjective characteristics of exposure to trauma; and (3) 22 items that provide an assessment of PTSD symptoms during the last month. Response options on the Likert scale are: never, rarely, sometimes, often, most of the time, with scores ranging from 0 to 4. The items evaluate the presence of intrusive memories, persistent avoidance, and persistent symptoms of increased excitability (distributed among B, C, and D criteria). They are evaluated by the severity criterion resulting from the sum of 17 items that meet the DSM-IV criteria (reaching a score of 38 or higher configures PTSD), proven by the sensitivity and specificity achieved according to Steinberg et al. [38–39]. This instrument was submitted to the cross-cultural adaptation process for the Brazilian adolescent population , with α = 0.866 for the items that compose the severity score (criteria B, C, and D) and significant correlations with the depression scale  (Spearman's rho = 0.405, p < 0.001). Good internal consistency was showed (Cronbach's α = 0.89).