thology is critical in adolescence because of the greater likelihood of earlier suicide attempts.

The most comprehensive longitudinal study of suicide pathways was conducted in New Zealand (Fergusson et al., 2000b). It focused on a cohort of 1265 children studied over the course of 21 years. The study sought to determine the extent to which social background, personality factors, mental illness, stressful life events, and childhood trauma contribute to suicide attempts.14 The childhood traumas (occurring before 16 years) were most of those covered by this chapter: sexual abuse, physical abuse, attachment to parents, caregiver separation/divorce or death, and parental substance abuse. Applying a proportional hazards model, three of the six predictors of suicide attempts at age 21 were related to childhood adversities: (1) child sexual abuse, (2) parental alcoholism, (3) low attachment to parents. The other three predictors were lower family SES and two child personality factors (neuroticism and novelty-seeking). The investigators then used a time dynamic model to account for the roles of mental illness and stressful life events in suicide attempts. This model found that none of the childhood traumas predicted suicide attempts independent of mental illness and stressful life events in adolescence. The study concluded that the effects of childhood traumas were completely mediated by mental illness and stressful life events. It suggested that the causal chain begins with childhood adversity, which increases the risk of suicide by increasing young people’s vulnerability to later mental health problems and stressful life events. In other words, both mental illness and exposure to stressful life events mediated the effect of childhood trauma on suicidality (Fergusson et al., 2000b).

The New Zealand study’s finding on child sexual abuse was not consistent with a finding from the U.S. National Comorbidity Survey. The major debate centered on whether psychopathology completely mediates the relationship between child sexual abuse and suicidal attempts, or whether child sexual abuse, by itself, without the presence of psychopathology, confers an independent risk (after controlling for confounding factors). These questions have important implications for prevention. If child sexual abuse is an independent predictor, then victims should be targeted for prevention programs, regardless of whether they have psychopathology.

In the United States study, Molnar and colleagues (2001a) found that, while the majority of suicide attempts were attributed to prior mental disorders, a significant percentage of suicide attempts occurred in the absence of psychopathology. The investigators suggested that methodological differences may explain the discrepancy with the New Zealand


Findings not presented for suicidal ideation.

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement