BACKGROUND/RATIONALE:
There has been extensive research indicating that psychiatric patients are at higher risk of suicide and accidental death than other patients, particularly those diagnosed with schizophrenia and depression. Other important risk factors include younger age, being male, abusing substances (particularly alcohol abuse), inadequate psychopharmacologic treatment, recent discharge from inpatient mental health programs, a past history of suicide attempts, and recent negative life events. Despite a plethora of studies on individual characteristics that are associated with suicide mortality in psychiatric patients, there has been no research on how quality of care might affect the risk of mortality in such patients. System-level characteristics that affect mortality are important to study since they are easier to modify than individual patient characteristics.

OBJECTIVE(S):
The three objectives are to: (1) examine geographical variation in the risk of suicide and other avoidable deaths among patients discharged from inpatient psychiatric programs; (2) determine whether standard quality of care measures are significantly associated with suicide or other external causes of death in the year after discharge; and (3) determine whether rates of suicide changed before and after the wide-spread closing of large numbers of VA inpatient mental health beds in 1996.

METHODS:
This study will have three parts, which will correspond to the three objectives. All of the parts will utilize VA administrative data to assemble samples, which will then be merged with National Death Index (NDI) data to determine mortality and causes of death. First, we will examine a sample of patients discharged from psychiatric bed sections between 1994 and 1998, the geographic variation in the rates of one-year mortality after discharge. Second, we will utilize administrative discharge and demographic data to examine what factors predict one-year mortality from suicide or accidental death after a psychiatric discharge from inpatient care. Third, we will compare those discharged before 1996 to those discharged after 1996 to examine whether reductions in inpatient beds influenced suicide rates.

FINDINGS/RESULTS:
Analyses of data indicate several things: 1) there are a number of individual factors that predict suicide mortality, including race, disability, diagnosis, length of stay, readmission and continuity of care; 2) there are few facility-level characteristics that predict mortality above and beyond individual characteristics; 3) there is substantial variation in suicide rates across VA facilities; 4) however, this variation is not explained by patient characteristics or by facility-level quality of care; and 5) social capital is one of the strongest predictors of suicide risk. Sub-analyses on suicide by firearm indicate that 1) local gun ownership rates are significantly associated with the risk of firearm suicide; 2) the restrictiveness of state gun laws are protective against the risk of firearm suicide; and 3) that high rates of social capital in the community are protective against firearm suicide.

IMPACT:
The effect of individual characteristics on risk of avoidable mortality is important, and several factors have been shown in these data to predict suicide mortality. However, healthcare systems can do little to change those characteristics. We explored the role of facility-level quality of care measures and found little evidence that differences in quality of care were associated with differences in suicide rates. These data suggest that suicide mortality may not be an adequate quality measure for the quality of health-care delivery at VA facilities. However, these data also suggest that clinical interventions to reduce access to firearms may help to reduce individual patients’ risk of suicide.

---

External Links for this Project

---

PUBLICATIONS:

None at this time.

DRA: Mental, Cognitive and Behavioral Disorders, Health Systems, Acute and Combat-Related Injury
DRE: Treatment - Observational
Keywords: Organizational issues, Safety
MeSH Terms: none