Suicide following presentation to hospital for non-fatal self-harm in the Multicentre Study of Self-harm: a long-term follow-up study

Background Self-harm is the strongest risk factor for subsequent suicide, but risk may vary. We compared the risk of suicide following hospital presentation for self-harm according to patient characteristics, method of self-harm, and variations in area-level socioeconomic deprivation, and estimated the incidence of suicide by time after hospital attendance. Methods In this ongoing Multicentre Study of Self-harm in England, the study population consists of individuals aged at least 15 years who had attended the emergency department of five general hospitals in Oxford, Manchester, and Derby after non-fatal self-harm between Jan 1, 2000, and Dec 31, 2013. Information on method of self-harm was obtained through systematic monitoring in hospitals. Level of socioeconomic deprivation was based on the Index of Multiple Deprivation (IMD) characterising the area where patients lived, grouping them according to IMD quintiles. Mortality follow-up was up to Dec 31, 2015, resulting in up to 16 years of follow-up. We calculated incidence of suicide since first hospital presentation by follow-up period and estimated the association between individual factors (age, gender, method of self-harm, IMD, and number of non-fatal self-harm presentations to hospital) and suicide using mixed-effect models. Findings Between Jan 1, 2000, and Dec 31, 2013, there were 92177 presentations to the study hospitals by 51108 individuals. 1325 patients involved in 1563 self-harm episodes were excluded from the study because they had missing information on gender, age, or mortality. The resulting study sample consisted of 90614 hospital presentations by 49783 individuals. By the end of follow-up on Dec 31, 2015, 703 patients had died by suicide. The overall incidence of suicide was 163.1 (95% CI 151- 5-175- 6) per 100000 person-years, and 260.0 (237.4-284.8) per 100000 personyears in men and 94.6 (83.3-107.4) per 100000 person-years in women. The incidence of suicide was highest in the year following discharge from hospital (511.1 [451. 7-578. 2] per 100 000 person-years), particularly in the first month (1787.1 [1423. 0-2244. 4] per 100000 person-years). Based on all presentations to hospital, men were three times more likely than women to die by suicide after self-harm (OR 3.36 [95% CI 2. 77-4 08], p<0.0001). Age was positively related to suicide risk in both genders, with a 3% increase in risk for every one-year increase in age at hospital presentation (OR 1.03 [1.03-1.04], p<0.0001). Relative to hospital presentations after self-poisoning alone, presentations involving both self-injury and self-poisoning were associated with higher suicide risk (adjusted OR 2.06 [95% CI 1- 42-2- 99], p<0.0001], as were presentations after self-injury alone (adjusted OR 1.36 11. 09-1- 701, p=0. 007). Similarly, relative to self-harm by self-poisoning alone, attempted hanging or asphyxiation (adjusted OR 2.70 [1- 53-4.78], p=0.001) and traffic-related acts of self-injury (adjusted OR 2.99 [117-7-65], p=0.022) were associated with greater risk of suicide. Self-cutting combined with self-poisoning was also associated with increased suicide risk (adjusted OR 1.36, [1- 08-1- 71], p=0.01). Compared with those patients living in the most deprived areas, those who lived in the least deprived areas (first national IMD quintile) had a greater risk of dying by suicide (adjusted OR 1.76 [1.32-2.34], p<0.0001) after adjusting for gender, age, previous self-harm, and psychiatric treatment, as did those living in the second least deprived areas (adjusted OR 1.64 [1.20-2.25], p=0.002). Interpretation Patients attending hospital for self-harm are at high risk of suicide, especially immediately after hospital attendance. Certain patient characteristics and methods of self-harm, together with living in areas of low socioeconomic deprivation, can increase patients' subsequent suicide risk. However, while specific risk factors can be usefully integrated into the assessment process, individual factors have poor utility in predicting suicide, so the needs and risks of all patients should be assessed to develop appropriate aftercare plan, including early follow-up. Copyright (C) 2019 Elsevier Ltd. All rights reserved.