Article Text
PDF
Abstract
Background and methods People with atrial fibrillation face an increased risk of thromboembolic events, and deciding on an antithromboembolic strategy in non-valvular atrial fibrillation is a prime opportunity for shared decision making. Therefore, tools facilitating this process are highly desirable. The American College of Cardiology strives for such with its ‘AnticoagEvaluator’, which is based on Peter Loewen’s ‘SPARCtool’. However, it appears these tools were released without standard peer review. Therefore, an analysis of these tools was undertaken to evaluate their soundness, namely proper use of baseline risk and effect estimates from the available evidence.
Results Despite Loewen’s laudable idea and the American College of Cardiology developing AnticoagEvaluator based on Loewen’s work, both tools have a flaw: they use relative effect estimates based on composite outcomes (thromboembolic and haemorrhagic) that do not match the baseline risk to which they are applied (thromboembolic). This can lead to importantly inaccurate impressions of therapeutic efficacy. This analysis explores this issue and offers potential solutions.
Conclusions The American College of Cardiology releasing and promoting a tool that gives misleading impressions of therapeutic efficacy is of considerable importance, though SPARCtool should also be corrected as a matter of importance. Means to correct the tools are identified herein, and if corrected, these tools stand to better fulfil their intended purpose as important and useful additions for clinical and shared decision making. This article’s analysis of the tools has a directly practical purpose, but it also serves as an instructive example of key elements of evidence-based medicine and shared decision making.
- biostatistics
- clinical epidemiology
- atrial fibrillation
- non-valvular atrial fibrillation
- evidence-based medicine
- shared decision-making
- decision aid
- direct oral anticoagulants
- novel oral anticoagulants
- warfarin
- vitamin K antagonists
Statistics from Altmetric.com
Footnotes
Contributors MM is the sole author and guarantor of this manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests Although I truly feel I have no conflicts of interest, in the interest of full disclosure, I declare the following: I conceived of and composed this manuscript when I was employed academically as a professor and clinically in internal medicine and cardiology. I remain clinically active, but I resigned my academic post at the end of June after being offered and accepting a position with EBSCO Health in their Innovations and Evidence-Based Medicine Development department. To whatever extent it might be considered important, I had no prospect of employment with EBSCO Health when I conceived of and composed this manuscript. My employment with EBSCO Health may seem relevant to some readers, because EBSCO Health is involved in, among other things, producing consumable resources for clinicians. As such, my critiques of AnticoagEvaluator and SPARCtool might be seen by some as potentially advantaging EBSCO Health by creating a perceived need for a more accurate tool. Importantly, however, and entirely irrespective of the timing aforementioned, the arguments I present herein are not subjective in nature, and as such, my employment is immaterial to my arguments. Furthermore, although I am critical of AnticoagEvaluator and SPARCtool in their present forms, I also suggest corrections and laud their efforts, and I am not advocating for the use of any other currently existing tool or any potential future tool, whether from EBSCO Health, any of its parent or subsidiary organisations, or any other entity entirely unrelated to EBSCO Health. I just want to contribute to evidence-based medicine and shared decision making being carried out judiciously.
Provenance and peer review Not commissioned; externally peer reviewed.
Patient consent for publication Not required.