Attachment, suicidal behavior, and self-harm in childhood and adolescence: a study of a cohort of Brazilian schoolchildren

Abstract

Abstract Background Attachment influences the development and the formation of the self and subjectivity and, just as early adverse events, may be related to the occurrence of mental disorders, suicidal behavior, and self-harm throughout life. This study aimed to analyze the effect of mental representation of attachment in children on suicidal behavior and self-harm throughout childhood and adolescence, considering the mediating role of internalizing problems. Methods Based on a cohort of 500 students (mean age 8 years, SD 1.2) sampled from public schools in a Brazilian southeastern metropolis, 316 children were followed for eight years in three waves (2006, 2008, 2012). The following data from the research baseline (2005) were used: family drawing, maternal and family variables, and sociodemographic data. The mental representation of attachment (independent variable) was measured by the Family Drawing Global Scale, discriminating between secure attachment and non-secure attachment. Suicidal behavior/self-harm (dependent variable) and internalizing problems were evaluated in three research waves through CBCL and YSR (ASEBA). Descriptive analysis, calculation of frequencies and p-values of the variables of interest, as well as modeling of structural equations, were performed. Results The prevalence throughout the study was: 17.1% [CI 13.3–20.8] for suicidal ideation and 8.9% [CI 5.6 – 12.2] for self-harm; there was a recurrence at one time in 16.5% [IC 12.6 – 20.3] and in two or more moments in 4.1% [CI 2.0 – 6.3] of the sample. Female gender (p = 0.035), internalizing disorders (p < 0.01), and non-secure attachment (p = 0.035) were associated with the occurrence of suicidal behavior/self-harm. The modeling indicated that 92,2% of the total effect of attachment (p = 0.069) on suicidal behavior/self-harm was due to direct effect, the other 7,8% of the effect being mediated by internalizing problems, adjusted for the confounding variables sex, skin color/race, and social stratum. The total effect showed a positive value, which indicates an increase in suicidal behavior/self-harm when the non-secure attachment is present. The approximate OR of non-secure attachment on the total effect (direct + indirect) was 1.15, indicating that, when adjusting for confounding variables, there was a 15% increase in suicidal behavior/self-injury from non-secure attachment. Conclusions The study supports the hypothesis that there is a relationship between disruptive attachment patterns (non-secure attachment) developed during infancy and suicidal and self-harm behavior during childhood and adolescence. These findings validate the concern about the first thousand days of childhood as a critical period for child growth and development, but also for the mental health of children and adolescents.