Background Post-stroke cognitive impairment (PSCI) is a frequent consequence of stroke, and reduces quality of life and increases care needs. We aimed to evaluate the impact of a hypothetical cognitive rehabilitation intervention on PSCI outcomes using the StrokeCog epidemiological model.

Methods We developed a probabilistic Markov model to project and track incidence and prevalence of PSCI in the Irish population aged 40–89 years to 2035. Data sources included official population and hospital episode statistics, and longitudinal cohort studies. Drawing on available systematic review evidence, we hypothesized that cognitive rehabilitation would reduce the risk of cognitive impairment no dementia (CIND) at 1 year post-stroke by 18% (scenario 1, S1, small effect) or by 54% (scenario 2, S2, medium effect) relative to usual care.

Results In usual care, the projected prevalence of post-stroke CIND in Ireland in 2035 was 6.7 per 1000 general population (95% CI 5.6–7.8), or 35% of stroke survivors (95% CI 30.5–38.8) (n=21026 prevalent cases). In S1 (small effect) the projected prevalence was reduced to 32.0% (95% CI 28.6–36.4) of stroke survivors (n=19652), and in S2 (medium effect) to 29.1% (95% CI 25.2–33.2) of stroke survivors (n=17672). The number of years of life lived free of cognitive impairment were increased by 6.3% in S1 (small effect) and 15.1% in S2 (medium effect).

Conclusion The StrokeCog model provides a tool for policy-makers and researchers to evaluate the potential impact of cognitive rehabilitation at different levels of intervention effectiveness. The model was based on conservative assumptions, and a less conservative approach could lead to a greater projected reduction in burden. Our next steps include analysis of quality of life outcomes and costs.