Cardiovascular diseases (CVD e.g. heart disease and stroke) are the leading cause of premature death and a major cause of disability. Doctors can use statistical models to predict people’s risk of developing CVD and decide whether to start preventative treatment.
Rheumatic diseases are diverse, including rheumatoid arthritis (RA), psoriatic arthritis, ankylosing spondylitis, for example. Inflammation is a key risk factor for CVDs, so these people have much higher risk. For example, people with psoriatic arthritis have 55% higher risk of CVD.
In RA, existing prediction tools have been shown to under-predict CVD risk, meaning people at risk may miss out on treatment. Despite this, risk prediction tools have not been tested for other rheumatic diseases (for this study defined to mean psoriatic arthritis (and psoriasis), ankylosing spondylitis, osteoarthritis and rheumatoid arthritis) to see how accurate they are.
Our first aim is to provide an updated description of how common CVDs are in people with rheumatic diseases in the UK, and whether this has changed over time with improved treatment options.
Second, we aim to test the accuracy of several existing general population risk prediction tools in people with rheumatic diseases. If these tools can be used for these people, their care will be improved by promoting use in clinical practice. If they are not valid for people with rheumatic diseases, our study will highlight the need for future research.
This study aims to assess cardiovascular disease (CVD) risk in patients with psoriatic arthritis, psoriasis, ankylosing spondylitis, rheumatoid arthritis, and osteoarthritis. We will select all patients within CPRD Aurum who have these conditions (HES APC will identify any additional individuals with the specified rheumatic conditions). We will identify which of these patients experienced a CVD event following diagnosis of their rheumatic disease using both CPRD Aurum and HES APC. The definition of CVD outcome differs slightly for each risk tool, but broadly speaking includes coronary heart disease, cerebro-vascular events, peripheral artery disease and heart failure.
Our project aims to (i) describe the 10-year incidence rates of CVD for each condition using segmented linear regression to assess whether this incidence has changed over time and with improved treatment and (ii) assess the performance of four general population CVD risk tools (QRisk3, Framingham risk score, Reynold’s risk score and SCORE) in patients with each condition. These risk tools underestimate CVD risk in patients with rheumatoid arthritis. We will determine whether this is also the case for other common rheumatic diseases.
Ten-year CVD risk will be calculated for each patient in our study using each of the four risk tools. The predicted risk will be compared to the observed risk. Discrimination will be assessed using area under receiver operating curve and calibration will be assessed.
CVDs are the leading cause of mortality in these patient groups. Improved understanding of their incidence (aim 1) will inform public health provision and prioritisation. It will also feed into patient-education material for lifestyle and self-management. Risk prediction and stratification (aim 2) will facilitate personalised prevention strategies, for example increased screening for CVDs and prevention using statins or lifestyle interventions in high-risk groups, optimising resource allocation for the health sector but also improve individual patient care and outcomes.
A primary outcome of cardiovascular disease will be measured according to the specific definitions in each of the risk scores to be examined.
1. The QRisk3 CVD risk prediction tool defines cardiovascular disease as a composite outcome of coronary heart disease, ischaemic stroke, or transient ischaemic attack.
2. The Framingham Risk Score defines cardiovascular disease as a composite of CHD (coronary death, myocardial infarction, coronary insufficiency, and angina), cerebro-vascular events (including ischemic stroke, hemorrhagic stroke, and transient ischemic attack), peripheral artery disease (intermittent claudication), and heart failure.
3. The Reynolds Risk Score defines cardiovascular disease as a composite of incident myocardial infarction, ischemic stroke, coronary re- vascularization, and cardiovascular deaths
4. The SCORE risk algorithm defines cardiovascular disease as a composite of ICD-9 codes 401 through 414 and 426 through 443, with the exception of the following ICD-9 codes for definitely non-atherosclerotic causes of death: 426.7, 429.0, 430.0, 432.1, 437.3, 437.4, and 437.5. We also classified 798.1 (instantaneous death) and 798.2 (death within 24 h of symptom onset) as cardiovascular deaths.
David Hughes - Chief Investigator - University of Liverpool
David Hughes - Corresponding Applicant - University of Liverpool
Jose Ignacio Cuitun Coronado - Collaborator - University of Liverpool
Kate Fleming - Collaborator - University of Liverpool
Sizheng Zhao - Collaborator - University of Liverpool
HES Admitted Patient Care;Patient Level Townsend Index