JA Gardner, CHQOER, Bedford, MA, VA Hospital; AM Hendricks, CHQOER, Bedford, MA, VA Hospital

Objectives: When the VA implemented the Veterans’ Equitable Resource Allocation (VERA) system to allocate the budget in FY1997, it created financial incentives for VA networks (VISNs) to change their mix of patients and the care delivered to them. VISNs whose budgets were reduced the most (losers) faced pressure to reduce resource use; those whose budgets were increased markedly (gainers) had less pressure to maintain efficiencies and more scope to expand resource use. This study examined inpatient care in all VISNs from FY1997 through FY1999 for five medical conditions to see whether resource use per patient treated (measured by length of stay (LOS)) changed across VISNs as expected under the new financial incentives.

Methods: A physician advisory panel identified five medical conditions for which there was substantial agreement about appropriate treatment practices (AMI, angioplasty, CABG, hip fracture, and the acute-stay portion of knee replacement). All VA patients with discharges in FY1997-FY1999 were identified by ICD-9-CM codes (primary diagnosis) or by ICD-9 procedure codes from the VA Patient Treatment File (PTF) and Extended Care File maintained at the Austin Automation Center (AAC). Cases in which the patient died were omitted. For each condition, linear regressions for LOS and log LOS estimated fixed effects of gainer/loser status of the treating hospital’s VISN under VERA, controlling for the patient’s age, general health status (ADG grouping or DCG score), economic conditions in the hospital’s MSA, the availability of community resources, specialization in the local medical care market, and the presence in the local market of special facilities or programs for treating the particular condition.

Results: For eacj condition, we found evidence of either increased resource use per patient for gainer VISNs or reduced resource use per patient for loser VISNs relative to other VISNs (neither gainers nor losers under VERA), as expected under the new financial incentives. But for no condition did we find expected changes in both gainer and loser VISNs. For example, for AMI, LOS fell in loser VISNs relative to other VISNs from no significant difference to about 6% shorter LOS ( p < .05). Similar reductions occurred for AMI and CABG. For knee replacement. average LOS increased in gainer VISNs relative to other VISNs, from 9% shorter LOS (p < .01) to 1.5% longer (not significant). For hip fractures, a similar increase occurred. For AMI, hip fractures and knee replacements, existing differentials in average LOS between gainer and loser VISNs widened in the expected direction (p < .05).

Conclusions: Differences and changes over time in LOS suggest that VERA has had an impact on resource use. However, we cannot measure the differences precisely enough, and the patterns we see are not exhibited consistently enough - either across medical conditions or among both VERA gainers and VERA losers - for us to declare confidently that VERA has had the effect on resource use that we would expect.

Impact: Information about the effects of VA’s budget changes on resource use and patient care is important for managers and decision makers in VA.