2055. Mental Health Service Delivery and Risk of Suicide in Psychiatric Patients
RA Desai, Northeast Program Evaluation Center, DJ Dausey, Northeast Program Evaluation Center, RA Rosenheck, Northeast Program Evaluation Center

Objectives: This study had four main objectives: first, to examine variation in suicide rates among discharged psychiatric patients across VA facilities; second, to examine individual and clinical factors that predicted suicide risk (i.e. case-mix); third, to examine individual delivery of care factors as they relate to suicide risk; and fourth, to examine facility and environmental factors as they relate to suicide risk.

Methods: All unique discharged psychiatric inpatients in the VA system between 1994 and 1998 with a diagnosis of PTSD, schizophrenia, major depression or bipolar disorder were included in the sample (n=121,933). The sample was merged with National Death Index data to determine death from suicide within one year of discharge. Administrative data on individual, clinical, service delivery (e.g. length of stay, continuity of care), facility-level variables (e.g. size of facility, academic affiliation, average length of stay), and social environment variables (e.g. social capital, State suicide rate) were used to build sequential GEE multivariate models.

Results: Out of 121,933 discharges, 481 patient committed suicide within one year of discharge (0.39%).Adjusting for case-mix (age, gender, race, disability, etc.), there was significant variation across VA facilities in the suicide rate. Service delivery variables that predicted suicide risk included length of inpatient stay, readmission within 180 days of discharge, and continuity of care.  No facility-level variables were significantly associated with suicide risk. Both social capital and the State-level suicide rate were significantly associated with suicide risk.

Conclusions: After adjusting for case mix, there are several individual delivery of care variables that, if modified, may affect the risk of suicide in psychiatric patients. However, there was no evidence that facility-wide characteristics or practice patterns were significantly associated with such risk. It may be that there are few system-wide policies that facilities could implement that would significantly impact on the risk for suicide among psychiatric patients.

Impact: These results could impact on individual clinical care, since there are several modifiable factors that predict suicide risk. However, these results also highlight the liklihood that there are few, if any, system-wide changes that would substantially impact this tragic outcome.