Medical guidelines recommend that people with more than a 1 in 10 chance of heart disease/stroke in the next 10 years should be offered long term treatment with medicines that may reduce that risk. These medicines are prescribed to very large numbers of people but are also associated with side effects. To focus treatment on people most likely to benefit, doctors use risk-prediction calculators to try to determine a person's chance of heart disease/stroke, but these calculators may perform differently in people with multiple health conditions (multimorbidity). The aim of this study is to examine how well current risk-prediction calculators for heart disease/stroke perform in people with multimorbidity, and whether they can be improved using different methods of analysis, in order to improve their accuracy for routine clinical practice.
Clinical guidelines recommend that patients with a cardiovascular risk of greater than 10% are treated with statins to prevent cardiovascular events. However, existing risk-prediction models may not account for the impact of multimorbidity and may be inaccurate in such patients. The aim of this study is to determine the accuracy of existing risk prediction models for cardiovascular disease (CVD) in people with multimorbidity. Existing risk prediction algorithms will be replicated in a cohort of adults without CVD using data on established risk factors. This cohort will then be used to derive and validate new risk prediction models accounting for competing risk in people with multimorbidity using a competing risk Cox proportional hazards regression model based on the method of Fine and Gray to estimate the associations between risk factors and outcomes. Discrimination will be examined using the area under the receiver operator characteristic curve by comparing the predicted CVD event rates from the model with observed rates for subsets of people with multimorbidity expected to have high competing risks of non-CVD death. The performance of these models will then be compared to existing models.
#NAME?
Bruce Guthrie - Chief Investigator - University of Edinburgh
Daniel Morales - Corresponding Applicant - University of Dundee
Peter Donnan - Collaborator - University of Dundee
SHONA LIVINGSTONE - Collaborator - University of Dundee
HES Admitted Patient Care;HES Admitted Patient Care;ONS Death Registration Data;ONS Death Registration Data;Patient Level Townsend Score;Patient Level Townsend Score