Background and Aims: Current cardiovascular disease (CVD) risk predictors, while advanced, fail to predict recurrent strokes or include significant age/ethnic groups and key modifiable factors. We aim to describe the methods to recalibrate and validate the Stroke Riskometer/PreventS-MD prediction algorithm for first-ever strokes, develop a new algorithm for recurrent strokes specific to NZ, and assess changes and disparities in CVD risk factor management in NZ general practices.
Methods: Employing a retrospective cohort study design, this research will analyse record-linkage data of NZ adults, aged 20-93 years, with and without a history of stroke/TIA. It will leverage national GP practice data linked with ARCOS datasets, CVD risk assessment datasets, and various national health databases to investigate stroke risk factors and management of cardiovascular factors for incident and recurrent strokes over the last two decades. Cox proportional hazards modelling will be used to develop new prediction equations for time to a first or recurrent stroke and first or recurrent heart attack/myocardial infarction using established modelling approaches, using pre-specified stroke and CVD risk variables. Predicted versus observed 5-year risk analysis and cross-cultural validation will be performed using Generalisability Theory.
Results and Conclusion: The study is expected to provide a recalibrated and validated prediction algorithm for first-ever strokes, a new algorithm for recurrent strokes, and insights into the changes and disparities in the management of CVD risk factors in NZ general practices that may enhance the effectiveness of primary and secondary stroke prevention, ultimately reducing the burden of stroke in NZ.