Can Novel Integrated Risk Schemes Change Treatment Decisions and Improve Health Outcomes in the Year after a Diagnosis of Atrial Fibrillation? An Individual-Based Simulation

Date of Approval: 
2016-08-01 00:00:00
Lay Summary: 
Patients with a common heart rhythm disorder called atrial fibrillation are at increased likelihood of stroke, which is caused by blood clots in the brain. To prevent the formation of these clots, physicians prescribe medicines called oral anticoagulants or “blood thinners.” All oral anticoagulants greatly reduce one’s chances of experiencing a stroke, while increasing one’s chances of a serious bleeding complication. In earlier work, we developed a set of prediction equations that physicians can use to calculate the probabilities of a patient first experiencing death, stroke, and bleeding in the year after atrial fibrillation diagnosis. Knowing these probabilities may help physicians choose treatment differently than they typically do. Different treatment may affect the chances of a patient experiencing stroke, bleeding, or dying and, as a consequence, the length and quality of life. The goal of this study is to compare health outcomes (stroke, bleeding, death, and health-related quality of life) that occur when physicians select treatment two different ways: as they typically do vs. using the prediction equations. To do this, we will conduct a computer experiment using data from a population of atrial fibrillation patients under various “what if” scenarios. The results from this study will guide future research.
Technical Summary: 
Treatment decisions in atrial fibrillation are often guided by risk stratification for stroke and bleeding, but the existing risk models have a number of limitations. We sought to address these limitations in an earlier study by developing new risk prediction models for the first competing events of stroke, intracranial hemorrhage, extracranial hemorrhage, and death. Our new models performed well in terms of discrimination and calibration. However, it is not known if our new models could change treatment decisions and impact subsequent health outcomes. The goal of this study is to evaluate the potential clinical usefulness of our new risk schemes using individual-based simulation. We will use data from a cohort of adult patients with incident atrial fibrillation to evaluate a number of treatment selection rules including current treatment patterns (patients are prescribed treatment we observed in the cohort). We will simulate health outcomes conditional on treatment and patient characteristics. Outcomes include stroke, intracranial hemorrhage, extracranial hemorrhage, death, death following an initially non-fatal event, all-cause deaths, and quality-adjusted life months. We will characterize the potential health impact of using our new models in treatment decisions through empirical confidence intervals about QALMs gained (primary) and mean events averted (secondary) versus current treatment patterns.
Health Outcomes to be Measured: 
The primary outcome will be quality-adjusted life months (QALMs). The secondary outcomes are the events that are used to derive QALMs: initial clinical events (stroke, intracranial hemorrhage, extracranial hemorrhage, and death), death within 30 days following an initial non-fatal clinical event, all-cause deaths.
Application Number: 
16_134
Collaborators: 

Morris Weinberger - Chief Investigator - University of North Carolina at Chapel Hill
Todd Durham - Corresponding Applicant - IQVIA - USA (Headquarters)
Anthony Viera - Collaborator - University of North Carolina at Chapel Hill
Jason Fine - Collaborator - University of North Carolina at Chapel Hill
Jayanti Mukherjee - Collaborator - Bristol-Myers Squibb - USA ( BMS )
Kristen Hasmiller Lich - Collaborator - University of North Carolina at Chapel Hill
Stacie Dusetzina - Collaborator - University of North Carolina at Chapel Hill

Linkages: 
HES Admitted Patient Care;ONS Death Registration Data