M Murdoch, Minneapolis VAMC Center for Chronic Disease Outcomes Research; K Halek, Minneapolis VAMC Center for Chronic Disease Outcomes Research; L Fortier, Minneapolis VAMC Center for Chronic Disease Outcomes Research; M Van Ryn, Minneapolis VAMC Center for Chronic Disease Outcomes REsearch; J Hodges, Univ MN School of Dentistry Oral Health Clinical Research Center and Univ MN School of Public Health Division of Biostatistics

Objectives: To assess the effect of a 1996,Veterans Benefits Administration (VBA) educational initiative to reduce such regional and gender-based discrepancies. Service connection is one of the most important predictors' of veterans' use of VA health care services and offers a "safety net" to otherwise uninsured veterans. Despite standardized protocols, the granting of service connection (SC) on the basis of PTSD has been previously shown to vary by gender and region.

Methods: Pre/post-intervention. Success rates of 1,218 men and 1,215 women filing disability claims for PTSD 2 years before the intervention were compared to the success rates of 1,242 men and 1,243 women filing claims 2 years after the intervention. Covariates potentially associated with a claim's success were retrospectively assessed through mailed surveys. The survey used validated measures to ask about veterans' PTSD symptomatology, in-service combat exposure and sexual trauma, post-service stressors, and physical and role functioning. Other data (e.g., service period and branch) were obtained through administrative databases.

Results: Overall, 3337 veterans returned surveys (69%). Response rates did not vary by pre or post-intervention status or by gender. Sociodemographic characteristics did not vary by pre/post-intervention status, but, except for race and income, men and women differed substantially.

The odds of becoming SC for PTSD increased in the post-intervention period for all veterans, but the interaction between gender and intervention period was non-significant (p = .71); that is, the VBA intervention did not affect the difference between men and women's claims approval rates. On crude analysis, the percent of women becoming SC for PTSD was 50% compared to 69% of men. Across VISN's, claims approval rates ranged from 44% to 72%. However, after adjustment for sociodemographic and service characteristics, PTSD symptomatology and physical functioning, in-service combat exposure and sexual trauma, and post-service stressors, the gender effect disappeared. The adjusted claims approval rate for women was 61% compared to 55% for men (p = .14). Regional differences remained after adjustment (p < .001), with adjusted claims approval rates ranging from 38% to 75% across VISN's. African American race and "cohabiting with a romantic interest" each had strong, negative effects on the odds of becoming service connected (both p's < .001). The adjusted claims approval rate for African Americans was 49% compared with 67% for all others and was 51% for respondents "cohabiting with a romantic interest" compared to 61% for other respondents.

Conclusions: Claims approval rates did not differ significantly by gender once important covariates were considered. Even after adjustment, however, veterans appeared almost twice as likely to become SC for PTSD at some VISN's than at others. African Americans and those "cohabiting with a romantic interest" were also less likely to be SC for PTSD.

Impact: SC is an important key to impoverished veterans' access to affordable health care. Differences in the odds of becoming SC for PTSD, as described here, could result in inequities and disproportionate hardships for African American veterans and for veterans seeking services in VISN's with low SC rates.