Rotation of retinal vascular arcades and comparison with disc-fovea angle in the assessment of cycloposition

Background/aims To assess the agreement between disc-fovea angle (DFA) and the retinal vascular arcades rotation to measure cycloposition. Device repeatability and reproducibility between observers were also evaluated. Methods Cycloplegic retinography was taken in 321 eyes of 165 normal patients and repeated in 18 eyes at least 24 h later. Two independent observers used software to determine DFA, vein-related angle (VRA) and artery-related angle (ARA) in every retinography. Mean value of related angle (MRA) (mean value of VRA and ARA) was calculated. Results Camera repeatability was good (Intra-class Correlation Coefficient, ICC 0.89). In Bland-Altman analysis, mean VRA, ARA and MRA were 4 degrees to 5 degrees different from DFA (p<0.01). There was poor correlation between DFA and other methods (ICC): DFA versus VRA 0.3, versus ARA 0.4, versus MRA 0.5. Reliability between observers was good in all methods. The multivariate analysis showed no interaction between each method and eye side, sex or observer (p>0.05). Conclusions The rotation of retinal vascular arcades using a first-order approximation technique is no substitute to DFA when assessing torsion in fundus photographs. Of the methods tested, MRA correlated most closely, but DFA remains the gold standard for cycloposition. The rotation of vascular arcades provides a qualitative assessment, particularly in uncertain macular location.