Interobserver Agreement of R2eD Brain Arteriovenous Malformation Scoring
Lea Scherschinski1, Christopher S Graffeo1, Stefan W Koester1
1Department of Neurosurgery, Barrow Neurological Institute, St. Joseph's Hospital and Medical Center, Phoenix, Arizona, USA.
Neurosurgery Practice
|June 5, 2026
Summary
Interobserver agreement for the R2eD arteriovenous malformation (AVM) score was low, particularly for nidus size. Standardization of R2eD AVM scoring is needed for reliable hemorrhage risk assessment in unruptured brain AVMs.
Area of Science:
- Neurosurgery
- Radiology
- Medical Imaging Analysis
Background:
- The R2eD arteriovenous malformation (AVM) score aids in assessing hemorrhage risk for unruptured brain AVMs.
- It supplements existing surgical risk scores and clinical decision-making tools.
Purpose of the Study:
- To evaluate the interobserver agreement among neurosurgeons using the R2eD AVM scoring system.
- To identify variability in scoring specific angiographic components of brain AVMs.
Main Methods:
- Three neurosurgeons independently scored 40 anonymized brain AVM imaging datasets.
- Scoring included R2eD components: location, nidus size, venous drainage, and arterial feeders.
- Fleiss kappa (κ) was used to assess interobserver agreement.
Main Results:
- Overall R2eD AVM score showed slight interobserver agreement (κ = 0.036, P = .52).
- Deep location demonstrated substantial agreement (κ = 0.689), while nidus size showed poor agreement (κ = -0.222).
- Venous drainage (κ = 0.400) and arterial feeder number (κ = 0.284) had fair agreement.
Conclusions:
- Significant interobserver variability exists in scoring individual R2eD AVM components.
- Further research is needed to address variability sources like tool familiarity and AVM complexity.
- Standardized R2eD AVM scoring is crucial for accurate hemorrhage risk stratification and clinical decisions.
