Stroke is the leading cause of long-term disability and a leading cause of death in the US. The VA has over 6,000 acute stroke admissions each year, and incurs direct costs exceeding $1 billion annually for care of new stroke patients and stroke survivors. Little is known about the composition of these costs, and their variation by VA Medical Centers (VAMC) and Veterans Integrated Service Networks (VISN).
The objective of this pilot study was to use comprehensive VA and Medicare administrative data (FY2001-10) to estimate the extent of geographic variation in healthcare utilization and costs for inpatient and outpatient care for patients hospitalized with acute ischemic stroke at a VA facility. Our aims are to 1) Estimate average patient-level risk-adjusted costs and healthcare utilization during the one year period following admission for acute ischemic stroke by VISN (FY2010); 2) Estimate temporal trends in costs and utilization by VAMC and VISN (FY2001-10); and 3) Examine longitudinal patterns in use of Fee Basis stroke care by VISN (FY2001-2010).
Using comprehensive VA inpatient data, we identified the primary study population consisting of admissions with a principal diagnosis of acute ischemic stroke at a VA facility (FY2001-2010). Treating the admission date for acute ischemic stroke as index date, we calculated healthcare utilization (acute and non-acute inpatient care, outpatient visits and outpatient pharmacy) and costs over the following 365 days ("post-stroke period"). To obtain a more accurate identification of healthcare and costs resulting from the stroke event, we also estimated healthcare utilization and costs during the 365 days preceding the index date ("pre-stroke period"). The difference in the pre- and post-period costs and utilization more accurately captures the impact of the acute stroke event. We captured costs of healthcare obtained within the VA and non-VA care covered by VA (Fee Basis) and Medicare. Comorbidity data was used to obtain risk-adjusted costs. We also used the clinically-detailed Office of Quality and Performance (OQP) Stroke Data (2007) to compare and validate the findings from administrative data.
1)As the largest component of one year total costs, we first examined index length of stay (days); given adequacy of sample size using FY2006-2010, we did not include FY2001-2005 so as to capture more recent trends. We grouped patients into those in the top 20% of length of stay ("long stayers"). Adjusting for patient risk and structural factors, we found that compared to the expected frequency of long stayers, the observed frequency was higher in 8 VISNs, lower in 9 VISNs and no different from expected in the remaining 4 VISNs.
2)Adjusted for pre-stroke costs, the median 1-year post-stroke costs were $19,916 (FY 2006-2010). Adjusting for patient risk and structural factors, we found that compared to the expected frequency of high cost patients (those in top 20 percentile), the observed frequency was higher in 7 VISNs, lower in 7 VISNs, and no different from expected in the remaining 7 VISNs. There was considerable overlap in the VISNs identified based on index length of stay and total costs.
3)Based on preliminary analysis we found that the number of Veterans hospitalized for stroke outside of VA (Fee Basis) was very small even in 2010, and therefore did not follow this line of enquiry.
Our goal was two-fold: first, to draw attention of VISN and VACO management and policymakers to the potentially large geographic variation in costs of post-stroke care broken down by the key components, and second, to identify structural factors that influence cost variation. Our study goals were consistent with "improving the value" of VA care.
None at this time.