Chest infections can trigger heart attacks and strokes. Currently doctors don’t generally consider the risks of heart attacks or strokes when treating chest infections. We think this may be a missed opportunity. There are two problems that need to be overcome: firstly we don’t know which patients with chest infections are most likely to have a heart attack or stroke. Secondly, even if we did know who was at risk, we don’t know how effective treatments would be at preventing the heart attacks and strokes.
This project will develop a computer based calculator that will help doctors predict the risk of heart attacks and strokes for patients with chest infections.
We will then look to see what the benefits and risks of treatments for these patients are. We will look at both aspirin and influenza vaccines, to see if they prevent heart attacks and strokes, and if there are side effects that could be damaging.
Aim –to investigate if aspirin and influenza vaccine, currently used for secondary prevention of cardiovascular disease (CVD), could be used as primary prevention for people at risk of infection-related CVD.
Population: adults over 40 years of age without prior CVD
Exposures: Respiratory infection, Aspirin, Influenza vaccine
Outcomes: Primary: CVD. Secondary: subtypes of CVD; bleeding/major bleeding
Objectives:
1. Derive a risk prediction model for infection-related CVD in CPRD Aurum.
2. Validate this prediction model in CPRD Gold.
For high-risk patients as identified by the model:
3. Examine the effects of aspirin on infection-related CVD.
4. Examine the effects of Influenza vaccine on infection-related CVD.
Methods:
1&2) We will use logistic regression to predict CVD in the 28 days following a first respiratory infection in CPRD Aurum and externally validate in Gold (using C statistics, E/O ratios, calibration plots).
3) To estimate the effect of aspirin on CVD, bleeding, and major bleeding in the 28 days following respiratory infection we will use logistic regression, instrumental variable (prescriber history, using ordinary least squares regression) and propensity score analyses (propensity modelled by mixed effects logistic regression). Sensitivity analyses are - timing of aspirin prescriptions, other antiplatelets (combined) and anticoagulants. Further analyses will examine instrument validity, magnitude and direction of bias.
4) We will examine the effect of influenza vaccine on CVD using Cox proportional hazards regression in a matched cohort. We will match vaccinated people to unvaccinated controls on age, sex, practice, date and indication for vaccination and follow them up for 180 days, until events, or the matched control is vaccinated. We will consider matching effective if there is no difference in infections between groups in the first 14 days, before vaccine protection starts. We will also use vaccine effectiveness for infection as an instrument.
Primary outcome: Cardiovascular disease events
Secondary outcomes: Coronary events; Cerebrovascular events; bleeding; major bleeding
Joseph Lee - Chief Investigator - University of Oxford
Joseph Lee - Corresponding Applicant - University of Oxford
Charlotte Warren-Gash - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Constantinos Koshiaris - Collaborator - University of Oxford
Cynthia Wright Drakesmith - Collaborator - University of Oxford
James Sheppard - Collaborator - University of Oxford
Jennifer Davidson - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Richard Hobbs - Collaborator - University of Oxford
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation