Methods: We identified patients with acute ischemic stroke in fiscal year 2001 who also had AF 3 years prior to the index stroke using ICD9 codes. Using a previously validated algorithm, we excluded patients with transient AF. Patients' inpatient and outpatient utilizations were collected from relevant VA and Medicare datasets. Patient's age, marital status, baseline comorbidities (hypertension, diabetes, heart failure, prior history of stroke, deep vein thrombosis, and prior bleeding), and distance to closest VA facility were compared between the three user groups. Geographic Information System technique was applied to visually describe and contrast the distribution of the three user groups.

Results: We identified 9,081 veterans with acute ischemic stroke of whom 13% had AF. We excluded patients with transient AF (n = 31) and/or those who died during their index hospitalization (n=129), leaving a final cohort of 1,030 patients, of whom 90% were VA-Medicare dually eligible. Among these 924 VA-Medicare dually eligibles, 74% received inpatient care within the first two years post-index stroke; 56% had VA only hospitalizations, 16% had Medicare only, and 28% had hospitalizations in VA and Medicare facilities. On average, Medicare only users were 2 years older than the other groups (76 vs. 74, p<0.05) and were less likely to have heart failure, diabetes, or bleeding. Similarly, 84% of the dually eligible patients used outpatient services within the two years post-stroke; 49% used VA only, 13% used Medicare only, and 38% used both VA and Medicare outpatient facilities. Distance to the closest VA inpatient/outpatient facility was 19.7/9.3 (VA only), 30.6/14.1 (Dual users), and 33.5/16.9 (Medicare only) miles, respectively (p<0.0001).

Conclusions: Most VA patients with stroke and AF were dually eligible patients who were more likely to use both inpatient and outpatient post stroke care in VA facilities. VA only users were located closer to both VA inpatient and outpatient facilities. Future work needs to define predictors of utilization for dually eligible users as well as examine differences in outcomes among user groups.