The finding: More than 34,000 Swedes (59% female, 41% male) who were hospitalized following intentional self-injury during 2000-2005 were followed for a mean of 5.3 years (range: 3-9 years) to assess risk factors for subsequent suicide completion. 1,182 individuals (57% male, 43% female) committed suicide during the follow-up period. In both men and women, the risk of subsequent completed suicide was highest in those with comorbid bipolar disorder; nonorganic psychotic disorder and moderate-to-severe depression were also associated with substantially heightened risk. In women, the presence of a personality disorder incurred a risk almost equivalent to that of a nonorganic psychotic disorder; this was not true in men, but the presence of any other affective disorder in men was associated with risk similar to that of a nonorganic psychotic disorder.
The study further assessed the method of self-harm during the index episode and noted that attempts involving active self-injury (hanging, strangulation, drowning, firearm, jumping, etc.) served as much stronger predictors of later suicide than attempts involving poisoning (overdose).
Combining these two observations identified high risk groups: 1 in 5 of those with bipolar disorder who used a means other than poisoning committed suicide within the follow-up period. Similarly, almost 1 in 6 of those with nonorganic psychotic disorders and 1 in 7 of those with moderate-to-severe depression whose index attempt involved deliberate self-injury by a means other than overdose completed suicide in the follow-up period.
The risk of subsequent suicide in individuals with anxiety disorders, substance use disorders, personality disorders, or other mental disorders was not negligible but this study underscores the amplified risk observed in those with bipolar disorder, psychotic disorder, and moderate-to-severe depression who also selected a method of initial self-harm that did not include poisoning. The same amplification of subsequent risk was not evident among patients with anxiety disorders, substance use disorders, personality disorders or other mental disorders who chose non-poisoning for their index attempt.

Strength and weaknesses: The Swedish National Patient Registry provides a comprehensive reference set. This study included only those individuals whose index episode warranted hospital admission. The prevalence of personality disorders may have been underestimated since this is a diagnosis that is not always fully appreciated or documented in the general hospital setting.

Relevance: The consultation psychiatrist regularly assesses patients who have been admitted to the general hospital following a suicide attempt. Inpatient psychiatric beds are not always readily accessible and available outpatient follow-up is sometimes tenuous at best, particularly for patients with limited resources. This study underscores the substantially elevated future risk of completed suicide for those with particular diagnoses who selected a means of self-injury other than overdose and thereby offers the consultation psychiatrist encouragement to pay particular attention to the disposition of these patients.