Deaths by Suicide and Other Causes Among Patients With Borderline Personality Disorder and Personality-Disordered Comparison Subjects Over 24 Years of Prospective Follow-Up.

OBJECTIVE: This study has 4 aims. The first is to determine rates of mortality due to suicide and other causes for patients with borderline personality disorder (BPD) and personality-disordered comparison subjects over 24 years of prospective follow-up. The second and third aims are to determine the best predictors of time-to-suicide and time-to-premature death (not due to suicide) in patients with BPD. A final aim is to determine whether mortality rates are impacted by recovery status.

METHODS: A total of 290 adult inpatients meeting rigorous Revised Diagnostic Interview for Borderlines and DSM-III-R criteria for BPD and 72 personality-disordered comparison subjects were recruited during inpatient admission at McLean Hospital between June 1992 and December 1995. Participants were followed and reassessed every 2 years, with data collection now entering its 26th year. Participant deaths were tracked over time.

RESULTS: A total of 5.9% of borderline patients and 1.4% of comparison subjects died by suicide. Additionally, 14.0% of borderline patients and 5.5% of comparison subjects died by non-suicide causes. Among borderline patients, number of prior hospitalizations significantly predicted completed suicide (HR = 1.62, P = .037). Sociodemographic factors, physical health indicators, and psychiatric history significantly predicted premature death (not due to suicide) in bivariate analyses (all P values < .05). In multivariate analyses, male sex (HR = 3.56, P = .003) and more prior psychiatric hospitalizations (HR = 2.93, P < .001) significantly predicted premature death. Most borderline patients who died either by suicide (87.5%) or non-suicide-related causes (88%) were not recovered before death.

CONCLUSIONS: Taken together, these findings suggest that individuals with BPD are at elevated risk of premature death. Patients who did not achieve recovery were at a disproportionately higher risk of early death than recovered patients.