QRISK®-3 is a tool that calculates a heart disease ‘score’ based on factors such age, ethnicity and whether a person has high blood pressure or diabetes. The score can predict how likely it is that someone will have a stroke or a heart attack for the first time in the next 10 years. Many General Practitioners (GPs) in the UK use the score during consultations with their patients. If their score is higher than recommended patients can be referred for further treatments to help reduce their risk of heart disease.
The QRISK®-3 score is very accurate, but the computer model only considers a small amount of information about reproduction and pregnancy. Including these features into the tool might help improve prediction of heart disease in women.
Our aim is to test whether adding reproductive and pregnancy-related features to QRISK®-3 helps make it better at predicting the risk of heart disease in women after childbirth. We will use data from primary care records to determine reproduction and pregnancy features and create computer models to decide whether adding these extra reproductive and pregnancy factors to QRISK-3 information helps improve the prediction of heart disease in women.
The revised tool will inform further research and help GPs make decisions about which women are more likely to develop heart disease after pregnancy. By finding out women who are most at risk of heart disease, treatments to reduce their risk can be accessed. This will benefit women, their families and the NHS.
The primary aim is to develop a robust 10-year risk prediction model for CVD in the postpartum period, which incorporates reproductive and pregnancy-related candidate predictors to aid clinicians, policymakers and patients in assessing the risk of future CVD postpartum. This will enable healthcare professionals to identify women at higher risk of CVD after childbirth so they can be referred to appropriate healthcare pathways.
We chose the postpartum period (the period after childbirth) because it has been identified in literature as a potential window of opportunity to prevent CVD in women [1].
We identified candidate predictors through a review of literature and conclusions from a clinical consensus meeting and Patient and Public Involvement (PPI) input. The candidate predictors were identified from three sources; i) the established QRISK®-3 CVD prediction model [2] ii) an umbrella review of reproductive risk factors of CVD in women [3], and iii) candidate predictors based on clinical knowledge from discussions within the study group.
The first step will be to externally validate the established QRISK®-3 CVD risk prediction model postpartum in pregnant women. Secondly, we will add reproductive and pregnancy-related features (identified from the umbrella review and consensus) to the QRISK®-3 equation to produce an updated prediction model. Finally, the performance of the updated model will be compared with that of QRISK-3 model.
The model will be fitted using Cox proportional hazards regression. Risk predictions will be evaluated at different time points (e.g., at 1 year, 5 years and 10 years), by producing calibration plots and estimating performance measures (discrimination and calibration), via internal validation.
Finally, the model will be independently and externally validated by separate researchers using the SAIL data. External validation will include assessment of discrimination (C-statistics), calibration (calibration plots and curves, calibration-in-the- large, calibration slope), and clinical utility (net benefit and decision curves).
First recorded diagnosis of cardiovascular disease (myocardial infarction, coronary heart disease, stroke, and transient ischaemic attack).
Krishnarajah Nirantharakumar - Chief Investigator - University of Birmingham
Steven Wambua - Corresponding Applicant - University of Birmingham
Anuradhaa Subramanian - Collaborator - University of Birmingham
Christopher Yau - Collaborator - University of Oxford
Francesca Crowe - Collaborator - University of Birmingham
Krishna Gokhale - Collaborator - University of Birmingham
Richard Riley - Collaborator - Keele University
HES Admitted Patient Care;ONS Death Registration Data;CPRD Aurum Pregnancy Register;CPRD GOLD Mother-Baby Link;CPRD GOLD Pregnancy Register;Patient Level Townsend Index