Geographic variation in poststroke depression among veterans with acute stroke

INTRODUCTION Depression after stroke is common. While some estimates range from as low as 12 percent to as high as 80 percent [1-2], most estimates indicate that poststroke depression (PSD) occurs among approximately 25 to 40 percent of stroke survivors [3-6]. Studies show that PSD negatively affects patients' functional recovery and quality of life and is associated with increased morbidity, mortality, and healthcare use [7-11]. Although PSD can be effectively treated with potentially life-saving outcomes [8,12-14], it is often underdiagnosed and undertreated [2,13]. Currently, no documentation is available regarding geographic variation in PSD detection. Understanding such variation could provide important clues for identifying patient and system factors that may contribute to the underdiagnosis and undertreatment of PSD. Significant geographic variation in stroke epidemiology and disparities in stroke incidence and prevalence have been demonstrated [15-19]. In a recent study of the 2005 Behavioral Risk Factor Surveillance System (BRFSS) survey, for example, overall stroke prevalence was reported as 2.6 percent among noninstitutionalized U.S. adults and the weighted, age-adjusted stroke prevalence varied across the states, with the lowest in Connecticut (1.5%) and the highest in Mississippi (4.3%) [20]. This overall estimate of stroke prevalence was comparable to previous study reports [16]. The BRFSS report showed that most states with high prevalence were clustered in the Southeast (SE) region, except for a few additional states in other regions (Illinois, Michigan, Missouri, Nevada, Texas, and West Virginia) that also had high stroke prevalence ([greater than or equal to]3.0%). This clustering of high stroke prevalence in the SE corresponded to the high stroke mortality rates previously observed in that region, which is traditionally known as the "stroke belt" [21-22]. No obvious simple explanation exists for the observed variation in stroke prevalence. Among the many different findings, the contributing factors to the variation included differences in the prevalence of stroke risk factors (e.g., high blood pressure, high blood cholesterol levels, diabetes, smoking, physical inactivity, and obesity) [18-19] and sociocultural and socioeconomic variations (e.g, cultural norms for diet and exercise, income, social isolation, and access to healthcare and preventable services) between the areas of high and low stroke prevalence [22]. Further studies addressing potential explanations for the stroke belt are still underway, including "The Reasons for Geographic and Racial Differences in Stroke Study" [19]. The Department of Veterans Affairs (VA), through the Veterans Health Administration, maintains the largest integrated healthcare delivery system in the United States and provides care to more than 5.2 million veterans each year at more than 1,300 sites. As such, VA medical databases provide unique resources for examining PSD detection and comparing geographic variation in PSD. However, many veterans covered by the VA healthcare system are also eligible for healthcare plans outside the VA system and many of them are dual- or triple-system users. For example, we recently found that among the 1,953 VA stroke inpatients living in the state of Florida, 30 percent were VA-only users, 60 percent were VA-Medicare dual users, 3 percent were VA-Medicaid dual users, and 7 percent were triple users within the first 12 months of their acute stroke hospitalization [23]. Consequently, obtaining veterans' PSD information from non-VA sources, such as other Federal and private plans, to gain a complete picture of their PSD status is important. Furthermore, VA provides its enrollees a comprehensive pharmacy benefit that could be a rich information source for identifying patients with PSD diagnosed outside the VA. The objective of this study was to use automated administrative data from multiple sources to compare geographic variation in PSD detection among a national sample of VA patients diagnosed with acute stroke across eight different geographic regions. …