Hospital management at index episode and relative risk of repetition within 12 months.

Statistically significant hazard ratios are highlighted in bold. These compare repetition in individuals receiving a particular aspect of management with repetition in all those not receiving that management (with the exception of ‘specialist community mental health follow-up’ where we exclude those with a psychiatric admission from the reference group).

Adjusted for baseline characteristics: main method of harm, drug/s used in self-poisoning (paracetamol/antidepressant/benzodiazepine), sex, age, ethnicity (White/Non-White/unknown), previous self-harm (yes/no/unknown), previous psych treatment (yes/no/unknown), current psych treatment (yes/no/unknown); standard errors and 95% CIs corrected for clustering by hospital.

The results for general hospital admission in Centre A are based on available data from a 5 year period, 2005 to 2009.