Atrial fibrillation (AF) is a health condition occurring when there are problems with the electrical signals in the heart causing it to beat irregularly. If left untreated, AF can damage the heart leading to serious health conditions such as stroke.
The risk of stroke in patients with AF can be reduced with drugs called anticoagulants, for example warfarin, which requires patients to achieve and maintain a level of blood clotting considered safe and effective. This requires regular monitoring at considerable expense to patients and the healthcare sector. Patient factors related to achieving good or bad levels of anticoagulation control are not adequately understood.
This study aims to use patient data routinely collected from general practices and hospitals to examine the relationship between the characteristics of patients with AF receiving warfarin (such as age, blood pressure etc.) and their level of anticoagulation control. This information will be used to explore whether patient factors can predict good or poor anticoagulation control, and thereby help to identify those that could benefit from alternative treatments to warfarin or better treatment management, thus reducing the chance of stroke. This study also aims to quantify the clinical and economic burden associated with poor anticoagulation control.
The effectiveness of warfarin as an anticoagulant is dependent on patients remaining within a narrow therapeutic range based on the International Normalised Ratio (INR). Patients receiving warfarin who are observed to not be stable within this range, or with high or low measurements, may require better management or switching to an alternative oral anticoagulant (OAC). However, outcomes could be improved if such instability could be accurately predicted at the time of anticoagulation initiation, before the patient receives warfarin.
The aim of this retrospective observational study is to evaluate empirical models that can be used in clinical practice for the prediction of anticoagulation control prior to warfarin initiation following AF. Using CPRD and linked HES data, various definitions of anticoagulation control will be explored, and the predictive performance of classical statistical models will be compared to that of machine learning techniques utilising baseline and time-varying covariates. Exploratory analyses will be used to explore clustering and temporal patterns in variables identified as risk factors for poor anticoagulation control. This study also aims to quantify the clinical and economic burden associated with poor anticoagulation control (incidence rates of all-cause mortality, clinical events and hospitalisations), using appropriate statistical techniques to adjust for observed confounders.
The primary health outcomes to be measured in this research protocol are:
- Percentage of time spent in the therapeutic range (TTR)
- Any of: two INR values greater than 5 or one INR value greater than 8 within a six-month period; and/or two INR values less than 1.5 within a six-month period; and/or TTR less than 65%
- <70% TTR
- Percentage of time spent under therapeutic range (TUR) - >/= 30% TUR
- Percentage of time spent over therapeutic range (TOR) - >/= 30% TOR
The secondary health outcomes to be measured in this research protocol are:
- Stroke
- All-cause mortality (defined using linked ONS mortality data)
- Stroke
- Transient ischaemic attack (TIA)
- Myocardial infarction
- Deep vein thrombosis
- Pulmonary embolism
- Bleeding event: intracranial, intra-articular, intracerebral, pericardial, gastrointestinal, intraocular, urinary, lung, other
- Bleeding event: intracranial, intra-articular, intracerebral, pericardial, gastrointestinal, intraocular, urinary, lung, other
- Hospitalisation: all admissions, and stratified by diagnosis for the above conditions where possible
- Length-of-stay: all admissions, and stratified by diagnosis for the above conditions where possible
- Healthcare resource costs associated with hospitalisations
Jason Gordon - Chief Investigator - Health Economics & Outcomes Research Ltd ( HEOR Ltd )
Jason Gordon - Corresponding Applicant - Health Economics & Outcomes Research Ltd ( HEOR Ltd )
Ameet Bakhai - Collaborator - Royal Free London NHS Foundation Trust
David Clifton - Collaborator - University of Oxford
Francisca Vargas Lopes - Collaborator - Pfizer Ltd - UK
Nathan Hill - Collaborator - Bristol-Myers Squibb Pharmaceuticals Limited - UK ( BMS )
Phil McEwan - Collaborator - Health Economics & Outcomes Research Ltd ( HEOR Ltd )
Steven Lister - Collaborator - Bristol-Myers Squibb Pharmaceuticals Limited - UK ( BMS )
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Townsend Score;Practice Level Index of Multiple Deprivation