Atrial fibrillation, a type of irregular heartbeat, poses a substantial risk for blood clots that enter the blood stream and cause strokes and heart-related diseases. To reduce the risk for brain attacks and heart diseases, patients with atrial fibrillation are prescribed anticoagulants, commonly known as ‘blood thinners’. Many types of available anticoagulants are effective for preventing complications. However, for patients just above the threshold at which anticoagulants are prescribed, treatment is contested due to its potential to cause bleedings. In addition, some families of drugs may not be used in patients with kidney disease, potentially leading to further kidney damage. Withholding drugs or reducing their doses to protect kidney function may also harm patients if these treatments are less effective. This creates a trade-off for clinical decision making. Clinical trials may not be fully able to capture real life effects occurring during routine care and may lack the long-run perspective to evaluate the benefits and potential risks of drugs. Studies that do not conduct experiments can take advantage of real-life clinical data but may fail to establish causality because many factors that are not controlled by the researcher may influence the results. To avoid both problems, this study uses electronic patient records. We focus on two types of patients with atrial fibrillation: First, we study patients with a borderline risk score at which treatment is prescribed. Second, we focus on the use of direct oral anticoagulants (DOACs) for patients with advanced kidney disease who are often excluded in clinical trials.
Anticoagulation is generally recommended above a CHA2DS2-VASc score ≥2 in men or ≥ 3 in women with atrial fibrillation. Treating patients with atrial fibrillation close to these thresholds is clinically contentious as the benefits from treatment (prevention of cardiovascular accidents) are relatively small compared to its risks (bleedings). For patients with severe renal impairment, some families of anticoagulants are contraindicated. As a result, physicians face a trade-off in clinical decision making because switching to other families of anticoagulants and withholding potentially effective treatments from patients in order to protect renal function may pose alternative health risks. Understanding the health effects of this trade-off is therefore critical for clinical practice. Here, we focus on different patient subpopulations: (i) with low CHA2DS2-VASc score and (ii) study the use of DOACs in patients with advanced kidney disease. Previous observational studies investigating the health effects of anticoagulants (and DOACs in particular) rely on identification of effects by adjusting for observable confounders but are unable to control for unobserved variables – and, therefore, cannot establish causality. Randomized controlled trials, in turn, might not be able to fully capture treatment effectiveness during routine care and frequently lack the time horizon to study long-term outcomes. By contrast, this study makes use of a regression discontinuity (RD) design, enabling the identification of causal effects in routine care with long-term outcomes. For this purpose, we draw on major clinical guidelines providing advice on the use of anticoagulants based on threshold rules related to the CHA2DS2-VASc score and patients’ glomerular filtration rate (GFR). Because physicians base their treatment decisions on considerations besides clinical guidelines, we use an instrumental variable approach that is robust to partial compliance. Our findings are expected to provide novel insights into the health effects of anticoagulants in patients with low CHA2DS2-VASc score and on DOACs therapy.
Primary outcomes:
To be analysed after five years follow-up from atrial fibrillation diagnosis:
- probability of an embolic event (stroke, renal infarction, splenic infarction, peripheral ischaemia) until follow-up
- probability of non-traumatic intracranial haemorrhage
- probability of requiring dialysis until follow-up
- number of GP visits until follow-up
- number of emergency hospitalizations (and their duration) until follow-up
- all-cause mortality until follow-up
Secondary outcome:
To be analysed after five years follow-up from atrial fibrillation diagnosis:
- estimated glomerular filtration rate (most recent before end of follow-up period)
- number of hospitalizations due to renal failure until follow-up
- mortality due to renal failure until follow-up
- risk of osteoporotic fractures
- dementia
- disability
Till Bärnighausen - Chief Investigator - University of Heidelberg
Maike Hohberg - Corresponding Applicant - University Hospital Heidelberg
Anant Jani - Collaborator - University of Oxford
Christian Bommer - Collaborator - University of Heidelberg
Duy Do - Collaborator - University of Heidelberg
Julia Lemp - Collaborator - University of Heidelberg
Justine Davies - Collaborator - University of Birmingham
Manuel Hoffmann - Collaborator - University of Heidelberg
Maximilian Schuessler - Collaborator - University of Heidelberg
Michaela Theilmann - Collaborator - University of Heidelberg
Pascal Geldsetzer - Collaborator - University of Heidelberg
Sebastian Vollmer - Collaborator - Georg-August-Universität Göttingen
2011 Rural-Urban Classification at LSOA level;HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation