Abstract
Purpose :
This study sought to illustrate where within-subject coefficient of variation (wCV) alone may be inconsistent when assessing repeatability and explore an alternative to wCV using ganglion cell complex (GCC) thickness and GCC focal loss volume (GCC-FLV).
Methods :
In this abstract, we assess repeatability for GCC thickness and GCC-FLV in normal eyes and eyes with pre-perimetric glaucoma (PPG) or perimetric glaucoma (PG). GCC-FLV integrates fraction loss of GCC in a significant loss mask. Mixed-effects models were used to calculate within-subject standard deviations, wCVs, and ICCs.
Results :
thickness (approximately 0.4% for all groups) than for GCC-FLV (>3% for all groups). Additionally, the wCV was more inconsistent across groups for GCC-FLV than for GCC thickness. The wCV for GCC-FLV was over 8.5% for the controls, more than twice that for PG subjects, likely due to the much smaller mean GCC-FLV for controls. In contrast, ICC was consistent across all three groups and for GCC thickness and GCC-FLV (all >0.99), consistently suggesting great repeatability.
Conclusions :
wCV was relatively large and inconsistent across groups for GCC-FLV. This was likely misleading as the values for ICC were all nearly 1. For GCC thickness, wCV was very small and consistent across all three groups. This was reflected in the values of ICC also being nearly 1. In conclusion, we recommend using ICC for metrics where the mean value is close to zero in order to assess repeatability as it is consistent than wCV.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.