Abstract
Purpose::
To determine if magnification correction using axial length (AL) or sphere equivalent (SE) refraction will improve glaucoma diagnosis.
Methods::
We analyzed a snapshot of the baseline data from the Advanced Imaging for Glaucoma study that consisted of 217 normal and 103 perimetric glaucoma (PG) eyes. The diagnostic parameters evaluated were retinal nerve fiber layer (RNFL) thickness measured by the Stratus OCT system (software Version 4.0) using the fast RNFL scan pattern. Magnification effects were measured by univariate regression of RNFL parameters against AL and SE in the normal group. The effect of age was also accounted for with a regression model. Generalized estimating equation was used to compensate for correlation between right and left eyes in the same person. The areas under the receiver operating characteristic (AROCs) were used to compare the glaucoma diagnostic performance before and after magnification or age adjustments. Weighted AROC analyses were also performed to compensate for the different distribution of age, AL and SE in the normal and PG groups.
Results::
Univariate linear mixed effect models of overall RNFL thickness against age, AL, and SE give R2 of 0.06, 0.09 and 0.02. The AROCs (value ± standard error) without magnification correction are 0.939 ± 0.014, 0.933 ± 0.015, and 0.89 ± 0.021 for RNFL overall, inferior and superior quadrant thickness averages, respectively. Magnification correction and age adjustment actually reduced AROC. The reduction was significant for some RNFL parameters. After equalizing the normal and PG groups with weighted AROC, we found that magnification and age adjustments did not significantly change diagnostic performance.
Conclusions::
Although magnification correction and age adjustments reduced the variance in the normal reference group, they did not improve the discrimination between normal and glaucomatous eyes. Part of the explanation is that glaucoma patients tended to be older and have longer eyes and more myopic refraction. But even after accounting for these biases, diagnostic performance was not improved by magnification and age adjustments. These results suggest that these adjustments may not be important in setting RNFL diagnostic criteria.Financial support: Supported in part by National Institute of Health contracts R01-EY13178-06, R01-EY013516-03 and P30-EY08098 (Bethesda, MD), The Eye and Ear Foundation (Pittsburgh, PA) and an unrestricted grant from Research to Prevent Blindness, Inc. (New York, NY)
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)