3118 — Comparing two methods for identifying non-veteran data in VHA utilization files
King SS (HERC), Shane AC
(HERC), Smith MW
To introduce two methods for identifying non-veteran data in the VHA utilization files. To illustrate how an apparent disparity in utilization between these two methods with respect to veteran status and gender reflects different rates of utilization by male and female veterans and non-veterans.
We examined VHA utilization in FY2002 by males and females using two different methods. We used the same variables to determine inpatient and outpatient statistics about non-veteran data. Method 1 obtained statistics about individuals, while Method 2 obtained statistics about encounter records. Corresponding statistics from these two methods were compared.
Method 1 found 359,617 inpatients (15,741 women and 343,876 men) and 4,436,298 outpatients (367,531 women and 4,068,767 men). Non-veteran inpatients accounted for 1.2% of all inpatients (15.2% of female inpatients) and 7.3% of all outpatients (51.7% of female outpatients). Over half (51.8%) of non-veteran outpatients were employees, and almost two-thirds (64.8%) of non-veteran outpatient employees were females. Method 2 found 563,215 inpatient admissions and 43,683,845 outpatient visits. Non-veteran encounter records accounted for only 0.9% of all inpatient admissions (12.8% of female inpatient admissions) and only 2.4% of all outpatient visits (24.5% of female outpatient visits). Almost half (46.5%) of the non-veteran outpatient visits were from employees, and almost two-thirds (63.4%) of the non-veteran outpatient employee visits were from females.
Studies about veterans, particularly female veterans, should consider eliminating non-veteran records, as these represent a substantial proportion of all care provided to females. The finding that non-veteran status is far more common in women than in men is much more apparent from Method 1 than Method 2. This is explained by the lower utilization of employees, who account for the majority of women non-veterans and who make minimal use of VHA.
VA researchers now have two methods to locate and, if desired, to eliminate non-veteran data. This will improve the accuracy of studies about the health care provided to veterans.