A lifetime economic model of mortality and secondary care use for patients discharged from hospital following acute stroke.

Background: The long-term health-economic consequences of acute stroke are typically extrapolated from short-term outcomes observed in different studies, using models based on assumptions about longer-term morbidity and mortality. Inconsistency in these assumptions and the methods of extrapolation can create difficulties when comparing estimates of lifetime cost-effectiveness of stroke care interventions.

Aims: To develop a long-term model consisting of a set of equations to estimate the lifetime effects of stroke care interventions to promote consistency in extrapolation of short-term outcomes.

Methods: Data about further admissions and mortality were provided for acute stroke patients discharged between 2013 and 2014 from a large English service. This was combined with data from UK life tables to create a set of parametric equations in a model that use age, sex, and modified Rankin Scores to predict the lifetime risk of mortality and secondary care resource utilization including ED attendances, non-elective admissions, and elective admissions. A cohort of 1509 (male 51%; mean age 74) stroke patients had median follow-up of 7 years and represented 7111 post-discharge patient years. A logistic model estimated mortality within 12 months of discharge, and a Gompertz model was used over the remainder of the lifetime. Hospital attendances were modeled using a Weibull distribution. Non-elective and elective bed days were both modeled using a log-logistic distribution.

Results: Mortality risk increased with age, dependency, and male sex. Although the overall pattern was similar for resource utilization, there were different variations according to dependency and gender for ED attendances and non-elective/elective admissions. For example, 65-year-old women with a mRS at discharge of 1 would gain an extra 6.75 life years compared to 65-year-old women with a mRS at discharge of 3. Over their lifetime, 65-year-old women with an mRS at discharge of 1 would experience 0.09 less ED attendances, 2.12 less non-elective bed days, and 1.28 additional elective bed days than 65-year-old women with an mRS at discharge of 3.

Conclusions: Using long-term follow-up publicly available data from a large clinical cohort, this new model promotes standardized extrapolation of key outcomes over the life course and potentially can improve the real-world accuracy and comparison of long-term cost-effectiveness estimates for stroke care interventions.

Data Assess Statement: Data are available upon reasonable request from third parties.