Stroke Prevalence, Incidence and Outcomes in Veterans with Diabetes
Principal Investigator: Thomas W. Findley, MD, PhD
Background: Stroke is a high impact disease in the VA, especially among individuals with diabetes who incur a 2 to 4 fold greater risk of stroke (Hart, 2000; Jamrozik, 2000); almost 40% of people with stroke also have diabetes (Manolio, 1996). Stroke is a key public health issue; recent studies conservatively estimate 750,000 stroke cases a year in the US (Williams 1999) of which 25 to 35% are recurrent (Ingall, 2000). Despite the reductions in incidence of stroke over the past 40 years, the reduction in mortality over the same time period means that even greater numbers of persons are surviving with sequelae of stroke (Findley 1987) However, more importantly, stroke is far more disabling than fatal, resulting in significant need for health care resources, loss of productivity, and decreases in quality of life.
Objectives: To use a unique longitudinal database of veterans with diabetes which links VA care with Medicare financed health care provided outside the VA to: (a) determine the incidence, prevalence and mortality of stroke in the veteran user population with diabetes; (b) determine process, functional status outcomes, and utilization in diabetic veterans admitted to VA rehabilitation centers (c) evaluate the preventive care provided before and after primary or secondary strokes using chart abstraction, controlling for factors of preventive care that are under control of the health care plan (region, access, utilization), those that may require care coordination (disability), and those risk factors that are not under the control of the plan (comorbid illness, age, and medication contraindications).
Methods: This project is a retrospective analysis utilizing the Diabetes Epidemiological Cohort (DEpiC), a database created for a separate, currently funded Merit Review Grant to study the epidemiology of diabetes in the VA (Drs Pogach and Miller, CO-Principal Investigators). A dataset for the proposed analysis has been derived from the DEpiC dataset. To take advantage of the availability of Medicare data from FY97-00, we have developed a FY97/FY98 diabetes cohort to model incident and subsequent stroke over 27months. Chart abstractions will be done for incident strokes in FY00 and FY01, when we are more likely to sample LHVS respondents and patients with a complete FIMÔ.
Findings/Results: "Stroke incidence among VA patients with diabetes" (under review) This paper addresses the inadequacy of ICD-9 code based strategies for identifying incident stroke in claims data. The ability to estimate the actual burden of stroke within claims data could assist prevention efforts for this costly disability, which is of particular import within the VA's aging diabetic population where stroke risk is greatest. Four algorithms (2 VA, 1 private sector, and 1 National Hospital Discharge survey based) were applied to matched VA and Medicare data from 1997 and 1998. Findings revealed shortcomings in each method and the need for further validation work. A chart review of 5000 records is planned.
Development of the chart abstraction tool has been undertaken by staff of EOVA and WVMI staff. The tool includes questions related to the research objectives: timing of stroke, stroke prevention interventions and education documented in the hospital setting or in outpatient treatment; patient behaviors related to diet, exercise and smoking, relevant medical diagnostic tests, laboratory and pharmacy information.
Status: Project work is ongoing.
Impact: Our findings provide new information regarding stroke incidence among veteran clinical users.