1. 1. To examine all deaths in patients aged 16–65 years in Nottingham who were known to secondary psychiatric care services between 1998–2003.

2. 2. To compare the numbers of suicides with numbers in the general population.

3. 3. To identify the predictors of suicide.

Ethical approval was granted by the local research ethics committee. Patients who died between 1998 to 2003 were identified from our electronic database. AS reviewed case notes and filled out data on a template.

The Office for National Statistics (ONS) provided the causes of death and coroners’ verdicts, and the total numbers of suicides in the general population. All data was coded, entered onto a spreadsheet and analyzed using SPSS version 15.0.

Analysis was conducted using T-tests and chi-squared analyses to differentiate between groups. Logistic regression was performed to identify predictors of suicide.

Using a logistic regression model, the variables that predicted suicide were age (p < 0.001), diagnosis of affective disorder (p < 0.001), previous self-harm (p < 0.001) and being admitted to a psychiatric unit in the last year (p = 0.028). For each increasing year of age upto 65 years, the odds of dying by suicide decreased by 11%.

In total, 275 people aged 16–65 committed suicide in the study period. Of these, 71 were open to psychiatric services at the time of death.

The majority of patients committing suicide are not known to psychiatric services. There is a need for stronger links with primary care to improve identification of people at risk of suicide.