Heart disease is the leading cause of disability and death in women. Previous studies have shown important differences in the start and advancement of heart disease in women compared to men. Nonetheless, the risk of heart disease remains poorly understood in women. Over the last ten years, the rates of pregnancy complications such as high blood pressure in pregnancy and high levels of sugar in the blood during pregnancy have steadily increased. Previous studies have shown that women with a history of complications of pregnancy are at higher risk of future heart disease. However, tools that are used currently by physicians to predict heart disease risk do not consider complications of pregnancy and the corresponding earlier onset of heart disease risk factors. Furthermore, these tools were developed over 20 years ago in older adult populations that are not representative of the general population or current clinical practice. There is therefore a need to develop valid model to predict heart disease that accounts for complications in pregnancy.
Although previous studies have shown an association between hypertensive disorders of pregnancy (HDP) and later CVD, the tools currently used by clinicians to predict a womens long-term risk of CVD do not account for complications in pregnancy such as HDP. Given the increasing rates of HDP and the limitations of previous risk prediction tools, there is a need to develop risk prediction models accounting for complications of pregnancy. However, the choice of inception cohort may impact the accuracy and performance of prediction models and also impact the generalizability of the models since women can contribute more than one pregnancy during the follow-up period. We will perform a plasmode simulation using the CPRD Pregnancy Register linked to Hospital Episode Statistics (HES) in order to determine how the inclusion of one versus multiple pregnancies per woman influences the accuracy and performance of risk prediction models. We will construct four cohorts based on a first delivery per women, a random sample of pregnancies, or all eligible pregnancies per women during the study period. Risk prediction models will be developed in each of these cohorts using women between the ages of 15-45 with a recorded delivery from April 1st, 2000 to March 31st, 2017. Deliveries will be identified from the Pregnancy Registry. The date of cohort entry will be defined as 42 days after the delivery date. Women will be followed until an event (incident cardiovascular disease) or censoring due to end of CPRD registration, last data collection, or end of the study period, whichever occurs first. Prediction models will be developed using an accelerated failure time models with time since cohort entry as the time axis. The accuracy and performance of models developed in each cohort will be compared.
Primary outcome: The primary outcome of incident CVD will be defined as, cerebrovascular disease, coronary artery disease, coronary revascularization, myocardial infarction, peripheral vascular disease, transient ischemic attack, unstable angina, or stroke
Samy Suissa - Chief Investigator - Sir Mortimer B Davis Jewish General Hospital
Robert Platt - Corresponding Applicant - McGill University
Graeme Smith - Collaborator - Queen's University Belfast
Jennifer Hutcheon - Collaborator - University Of British Columbia
Kristian Filion - Collaborator - McGill University
pauline reynier - Collaborator - Sir Mortimer B Davis Jewish General Hospital
Sonia Grandi - Collaborator - McGill University
HES Admitted Patient Care;ONS Death Registration Data;Practice Level Index of Multiple Deprivation;Pregnancy Register