Lack of concordance between empirical scores and physician assessments of stroke and bleeding risk in atrial...

Area of Science:

• Cardiology
• Clinical Medicine
• Health Services Research

Background:

• Atrial fibrillation (AF) management requires balancing stroke prevention with bleeding risk.
• Empirical risk prediction models exist, but their alignment with physician assessments is not well-understood.

Purpose of the Study:

• To compare physician risk assessments with empirical scores for stroke and bleeding in AF patients.
• To investigate factors influencing physician risk assessment compared to established models.

Main Methods:

• Analysis of 10,094 AF patients from the ORBIT-AF registry.
• Comparison of CHADS2 (stroke) and ATRIA (bleeding) scores with physician-assigned risk categories (low, intermediate, high).

Main Results:

• Low agreement observed between empirical scores and physician assessments for both stroke (weighted Kappa 0.1) and bleeding risk (weighted Kappa 0.11).
• Physicians underestimated high stroke risk (16% vs. 72% by CHADS2) and bleeding risk (7% vs. 17% by ATRIA).
• Anticoagulation use correlated with high stroke risk but was minimally impacted by bleeding risk assessments.

Conclusions:

• Significant discordance exists between provider-assessed and empirical risk scores in AF.
• This disagreement may partly explain why anticoagulation choices deviate from clinical guidelines.