More people are surviving and living with coronary heart disease (3 million people in the UK). Some go on to have further heart problems after their first heart disease event and die prematurely, while others do not. Despite our knowledge about what causes heart disease and how to prevent it (smoking cessation, diet, exercise), we know little about factors that affect the chances of progression to a second event - which may be different. As a result we can't tell who is at higher risk for further problems and are limited to treating the usual risk factors we know about for first events, with little evidence supporting this approach.
In this proposal we plan to use anonymous GP and hospital records to (1) see whether our usual cardiovascular risk factors are still important in people once they have developed heart disease (2) which other risk factors and diseases are relevant for disease progression (3) whether any existing drugs we already use for other conditions have any useful effects for this group and (4) whether we can combine this information into a risk score to try and better identify those at greatest risk of recurrent or subsequent heart problems.
Among patients surviving their first coronary heart disease (CHD) event, risk of further cardiovascular events remains high. Modifiers of subsequent event risk are poorly understood, compared to those for a first CHD event in the general population. Understanding these modifiers has important implications for secondary prevention of CHD.
We hypothesise that (1) traditional cardiovascular risk factors may have different relationships for first and subsequent CHD events (2) there are novel and specific risk factors relevant for subsequent CHD events (3) non-cardiac drugs may have unrecognized benefits or harms for subsequent event risk and (4) this information could be combined to generate better risk modelling for use in patients with established CHD.
Using linked electronic health records, we will use Cox proportional hazards to model the associations between risk factors and first CHD and subsequent CHD events. Further Cox models will be used to estimate the associations between non-cardiac drugs and risk of subsequent CHD events, adjusted for propensity scores which model probability of treatment to limit confounding by indication bias. Multivariable logistic regression will be used to develop risk prediction models for subsequent events. In validation, C-indexes and Hosmer-Lemeshow tests will be used to assess the model discrimination and calibration.
Health Outcomes to be Measured:
- Unheralded coronary death
- Peripheral arterial disease
- Coronary revascularisation
- Heart failure
- Transient ischaemic attack
- Unstable angina
- Intracerebral haemorrhage
- Subarachnoid haemorrhage
- Ventricular arrhythmia or sudden cardiac death
- Peripheral arterial disease
- Abdominal aortic aneurysm
- Ischaemic or unspecified stroke
- All-cause mortality
- CHD cause of death
- CVD cause of death
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation