Introduction: Reductions in early case-mortality after stroke have occurred over the past few decades, but whether this has translated into increases in longer term survival is unknown.
Hypothesis: Case-mortality at 1 and 5 years after stroke will have decreased over time.
Methods: Individual participant data on sociodemographics, pre-stroke dependency, stroke type and severity (e.g. NIH Stroke Scale, Glasgow Coma Scale) and vital status (including date of death) were obtained from study investigators from 12 high-quality population-based stroke incidence studies undertaken in Australasia, Europe, South America and the Caribbean between 1987 and 2012. We used random effects Poisson regression to model changes in case-mortality at 1 and 5 years after stroke between 1987 and 2013, adjusted for age, sex and stroke severity.
Results: There were 16,557 participants with 1 year follow-up (mean [SD] age 73.6  years; 53% female, 79% ischaemic stroke, 21% intracerebral hemorrhage) and 12,839 with 5 year follow-up. In 1987, case-mortality was 42.9% at 1 year and 61.5% at 5-years. By the end of the study period, 1-year case-mortality was 31.3% and 5-year case-mortality was 49.6%. During the period 1987 to 2013 adjusted 1-year case-mortality following stroke decreased by 0.56% (95% CI 0.55, to 0.57) per year independent of age, sex and stroke severity (Figure, black line). Similarly, the adjusted 5-year case-mortality decreased by 1.19% (95% CI 1.17 to 1.20) per year (Figure, blue line).
Conclusion: Long term survival after stroke has improved since 1987. Further work is required to determine the relative role of changes in natural history, secondary prevention and clinical care on these findings.
Figure. Reduction in 1 year (12 studies during 1987-2013, n=16,557) and 5 year (7 studies during 1987-2008, n=12,839) case-mortality according to year of stroke estimated with random effects Poisson regression.