Cardiovascular disease (CVD), comprising mostly heart attacks and strokes, is UK's leading cause of death and disability. Measurements of "risk factors" that are linked with greater risk of CVD (e.g., older age, male sex, smoking, high blood pressure, high cholesterol) are often recorded in electronic health records during visits with doctors. Information on these risk factors that are already recorded in electronic health records can be used to estimate how likely someone is to develop CVD over the next ten years and allow doctors to invite people with a high risk in for a full assessment of their risk. This approach could help identify and treat people at risk of developing heart disease earlier, with the ultimate aim of reducing the number of CVD events and deaths. Historical information on a person's risk factors can also be used to produce personalised estimates of when a person should next visit their GP for a CVD risk assessment. This project will use information from medical records from CPRD to: 1) evaluate and develop CVD risk prediction models that can be applied to health records to flag people at high risk of developing heart disease; 2) investigate the impact of different risk thresholds for inviting people in for formal risk assessment; and 3) develop methods for using medical records to provide personalised recommendations on frequency and timing of CVD risk assessments.
Stratification of individuals according to their estimated cardiovascular disease (CVD) risk is used to guide clinical decision-making. Current UK guidelines for CVD risk assessment recommend the use of already recorded risk factors in electronic health records to prioritise patients for a full formal risk assessment, although there is no guidance on how this should be achieved. Our team is currently developing methods for a CVD risk modelling approach that utilises already available information on risk predictors while accounting for sporadically observed/missing data and can estimate risk for a large proportion of the target population using electronic patient records from a subset of 10 general practices contributing to The Health Improvement Network (THIN10). The proposed work in CPRD will involve several follow-up avenues to this work including: applying these methods to derive and internally validate a risk prediction model using a large sample size from CPRD with validated CVD endpoints and undertake public health modelling, optimising thresholds for a CVD "pre-screening tool", and methods work investigating optimal personalised and stratified (i.e. stratified by risk thresholds) screening intervals.
10-year risk of developing fatal or non-fatal cardiovascular disease
Angela Wood - Chief Investigator - University of Cambridge
Angela Wood - Corresponding Applicant - University of Cambridge
David Stevens - Collaborator - University of Cambridge
Ellie Paige - Collaborator - The Australian National University
Jessica Barrett - Collaborator - University of Cambridge
Juliet Usher-Smith - Collaborator - University of Cambridge
Luanluan Sun - Collaborator - University of Cambridge
Matthew Arnold - Collaborator - Astra Zeneca Ltd - UK Headquarters
Robson Machado - Collaborator - University of Cambridge
Stephen Kaptoge - Collaborator - University of Cambridge