Abstract
The objective of this study was to investigate risk factors that could help to differentiate between 14–25-year-old suicide attempters who survive the attempt and those who do not. Using a case–control psychological autopsy method, 29 cases (completers) and 23 controls (survivors) matched for age and sex were included. Survivors, parents and a close friend participated independently in semistructured interviews. Kiddie-Schedule for Affective Disorders and Schizophrenia Lifetime, stress scales, medical records and a sociodemographic inventory were used to reach a consensus best estimate DSM-IV five-axis diagnosis for each subject. A conditional logistic regression was used with the PHREG procedure in SAS to investigate the relationship between the binary outcome (survival/death) and the set of putative risk factors. No significant Axis I diagnostic differences were found between groups. Based on odds ratios, school suspension and irritability provide the best predictive factors for survival. Together with an Axis II cluster C diagnosis (avoidant, dependent and obsessive-compulsive personality disorder), high Global Assessment of Functioning scale, reactivity of depressed mood and level of obsessions were associated significantly with suicide completers. Cases appear to show more of an ‘internalizer’ profile, compared to controls who were more of the ‘externalizer’ profile. Dissatisfaction with physical appearance, rejection from others and feelings of isolation were highly prevalent in all subjects. Seventy-two per cent of cases died on their first attempt. The vast majority of subjects sought general medical care in the months preceding the suicidal behaviour. A mental health professional was consulted within the previous year by a greater proportion of controls, compared with cases. Our research emphasizes that the combination of psychopathology (diagnosed or not) and previous high level of functioning (in individuals who internalize their conflicts) represents an at-risk group for youth completing suicide. These findings underscore the importance of mental health evaluation as part of the general medical care assessment.
Acknowledgements
With appreciation, we recognize the support and funding from Fondation Pierre-Janet (PR 93-05-01B), the Royal Ottawa Hospital Associates in Psychiatry and the CHEO Research Institute (93-32S (E); cc9898). This work is dedicated to all our participants who allowed us to come so close into their lives.