1005 — Using a Comprehensive Dataset to Understand the Characteristics, Utilization, Efficiency, and Outcomes of VA Patients' VA and Non-VA Inpatient Care
Weeks WB (VA Outcomes Group, Research Enhancement Award Program (REAP)), West AN
(VA Outcomes Group, REAP), Wallace AE
(VA Outcomes Group, REAP), Fisher ES
(VA Outcomes Group, REAP)
VA currently uses a VA/Medicare dataset to determine VA enrollees' reliance on the VA for inpatient care. However, most VA enrollees who are younger than age 65 are not concurrently enrolled in Medicare. Therefore, we sought to develop and evaluate a comprehensive inpatient dataset of VA and non-VA inpatient care provided to VA enrollees for both younger and older VA enrollees in a large state, New York.
Using VA and New York State administrative and clinical databases, we used probabilistic matching algorithms to develop a dataset of inpatient admissions for 110,716 residents of New York State who were enrolled in VA healthcare and had 266,869 inpatient admissions in the state in 1998-2000, either within or outside of the VA system. For each admission, we collected information on the system of care used (VA or non-VA) and admission characteristics. We compared reliance, admission characteristics, and risk-adjusted outcomes for VA to non-VA care for two groups: veterans who were younger than age 65 and those age 65 or older.
About 50% of younger patients' inpatient admissions were in the VA compared to about 30% of older patients'; however, reliance on the VA varied dramatically across the 19 Major Diagnostic Categories (MDCs) examined. In either age group, those admitted to the VA were younger, less likely to be white, and less likely to live in a rural setting than those obtaining non-VA care. Patients using the VA had lower Charlson scores and received less complex care. For both age groups and across all MDCs, VA admissions had substantially higher DRG-specific observed-to-expected lengths-of-stay. Compared to non-VA admissions, VA admissions had longer risk-adjusted lengths-of-stay and higher 30-day mortality rates. By meeting local care standards, VA's required bed-days-of-care would decrease by 49%.
In New York state, compared to the private sector, VA appears to provide less efficient inpatient care that is associated with worse outcomes.
Datasets like the one we developed could inform planners, researchers, clinicians, and managers about VA's service market, including both older and younger VA enrollees' reliance on VA. Such datasets could be used to develop achievable efficiency and performance benchmarks.