JE Zeber, SMITREC; L Gillon, SMITREC; M Valenstein, HSR&D COE Ann Arbor/SMITREC; JF McCarthy, SMITREC; FC Blow, SMITREC

Objectives: Global Assessment of Functioning (GAF) scores are widely used as an appropriate and efficient indicator of functional status. The measure has excellent construct validity, with strong correlation with the condition and symptomatology of seriously mentally ill patients. Per a 1997 VA directive (#97-509), every veteran in a mental health program should be administered the GAF at least quarterly for outpatient care and upon hospital discharge. However, little research has explored the relationship between GAF scores and ethnicity, potential systematic bias in score reporting, or the implication of either differential or missing scores on the quality of patient care.

In addition to monitoring compliance with the 1997 directive, this study examines differential reporting of GAF results between ethnic groups (African-Americans vs. Caucasian), in terms of both actual score disparity and the prevalence of missing scores. The degree of variation among VISNs on these measures will also be explored.

Methods: Using data from the Patient Psychoses Registry, veterans with a diagnosis of schizophrenia, bipolar disorder, or 'other' psychoses during fiscal year 1999 were identified through the Patient Treatment File (PTF) or Outpatient Care Files (OPC). Inpatient admission and outpatient GAF scores (1-100) were analyzed after controlling for age and gender. A case-mix adjustment, based on psychiatric outpatient visits and inpatient days, was used to control for the severity of patient conditions. T-tests for mean differences are reported for overall outpatient and inpatient scores, plus separate means by diagnoses. To compare missing GAF scores between ethnic groups, logistic regressions were run.

Results: The GAF scores of 150,887 SMI veterans (black=37,757, white=113,130) were examined. The average inpatient GAF for black schizophrenics was higher than whites (40.60 vs. 39.35, p=.0001), but the outpatient average GAF score favored whites for all diagnoses (51.08 vs. 50.77, p=.0157). None of the GAF score differentials reached clinical significance (i.e. 3 to 5 points), perhaps a more important finding regarding assessment and treatment. However, a notable disparity in 'missing' scores was observed across treatment settings. Interestingly, although blacks were 19-28% less likely to be assessed on the outpatient GAF, they had a 16-19% greater chance of receiving inpatient scores. Compliance with the VA directive was 50.9% for inpatient scores and 58.6% for outpatient. Finally, preliminary observations regarding variation across the system are promising and warrant further study.

Conclusions: Despite a commitment to assessing the functional status of all mentally ill veterans, only one-half of all patients had an appropriately recorded GAF score in FY99. Equally troubling, small but striking disparities exist between ethnic groups, depending on diagnosis and/or treatment setting. Although this study does not specifically examine issues surrounding the severity of illness, utilization or outcomes, inconsistent and potential bias could signal differential assessment and treatment of minority veterans with serious mental illness.

Impact: These data highlight the degree of compliance with the VA's policy on GAF reporting, and emphasize the need to target especially vulnerable patients with serious mental illness.