Repetition risk for adolescent self-poisoning: a multiple event survival analysis

To examine risk factors associated with re-presentation (event rates) following an initial episode of hospital treated self-poisoning.

A longitudinal cohort study using the Hunter Area Toxicology Service (HATS) database of all presentations to hospital of self-poisoned patients aged 10-19 in Newcastle and Lake Macquarie Regions of New South Wales from January 1991 to December 1995. The study factors were: age, gender, employment status, 'substance abuse' and psychiatric diagnosis at index (first documented episode during the study time-period) admission. The main outcome measure was re-presentations per unit time. Time-event analysis (multivariate) was used to compare re-presentation rates per person-year exposure to the study factors.

There were 450 patients who presented on a total of 551 occasions. The median and modal age at initial presentation was 17. Three hundred and nine (69%) were female and 141 (31%) were male. The probability (95% CI) of a patient re-presenting within one year of an index admission with self poisoning was 0.09 (0.07-0.12) and within 5 years was 0.16 (0.12-0.21). The adjusted rate ratios for episodes of re-presentation were: any 'substance abuse (ever)' 3.87 (2.08-7.21), 'alcohol abuse' 2.32 (1.15-4.68),'benzodiazepine abuse' 4.89 (1.63-14.62), schizophrenia and other psychotic disorders (DSM-IV) 2.85 (1.2-6.79), and any personality disorder (DSM-IV) 2.68 (1.73-4.16).

Interventions to decrease recurrence rates for adolescent self poisoning should be directed towards substance (particularly alcohol or benzodiazepine) abuse, non-affective psychoses and personality disorder.