Risk of long-term post-stroke dementia using a linked dataset of patients with ischemic stroke without a history of dementia

Background: Post-stroke dementia (PSD) is a common and disabling sequela of stroke. However, the long-term incidence of PSD after an ischemic stroke and factors which predict its occurrence are incompletely understood. Linkage of large health datasets is being increasing used to study long-term outcomes after disease. We used large-scale linked data from Korea to determine the long-term incidence of PSD after ischemic stroke, and identify which factors predicted its occurrence. Methods: From January 2008 to December 2014, patients with ischemic stroke ( n = 37,553) without a history of dementia were included in a linked dataset comprising the claims database of the Health Insurance Review and Assessment Service and the Clinical Research Center for Stroke registry data. The outcome measure was PSD after ischemic stroke. Clinical factors evaluated included vascular risk factors, acute stroke management including reperfusion therapy, antithrombotics, and statins, stroke severity, and educational levels, were evaluated. Results: Among 37,553 patients with ischemic stroke without a history of dementia (mean age: 64.9 years; 61.9% males), 6052 (16.1%) experienced PSD during a median follow-up period of 5 (interquartile range, 3.4–7.0) years. The 10-year estimated cumulative incidence of dementia was 23.5%. Age (hazard ratio (HR) 1.82 per 10 years, 95% confidence interval (CI) 1.75–1.88) and a lower educational level (illiteracy or no education HR 1.65 (CI = 1.44–1.88), 0–3 years 1.53 (CI = 1.31–1.79), 4–6 years 1.60 (CI = 1.43–1.80), 7–9 years 1.32 (CI = 1.16–1.49), 10–12 years 1.17 (CI = 1.04–1.32)) were independently associated with an elevated risk of PSD. Male sex was associated with a significantly lower risk of PSD (HR 0.86, CI = 0.79–0.92). Diabetes mellitus (HR 1.21, CI = 1.14–1.29), a history of stroke before index stroke (HR 1.31, CI = 1.21–1.41), and initial National Institutes of Health Stroke Scale (HR 1.03, CI = 1.03–1.04) were independent risk factors for PSD. Regarding medications, the use of anticoagulation and antipsychotic medications after stroke appeared to be associated with increased PSD risk, whereas statin therapy was associated with a reduced risk. Conclusions: PSD is common with a 5- and 10-year incidence in patients with ischemic stroke without a history of dementia of 16.1% and 23.5%, respectively. Factors associated with PSD include age, female sex, lower educational level, diabetes mellitus, initial stroke severity, antipsychotics, and anticoagulants. Further studies are required to determine whether reducing those risk factors which are treatable reduces the incidence of PSD.