Abstract
Methods: :
Consecutive participants in the population-based Rotterdam Study were examined with both HRT and simultaneous stereoscopic fundus photography in addition to other ophthalmic examinations including visual field testing. Normative values for all HRT3 variables were determined in participants without glaucomatous visual field loss (GVFL). Sensitivity (as determined in participants with GVFL) at a fixed high specificity of 0.975 and agreement with GON as previously defined using simultaneous stereoscopic fundus photography were taken into account for the HRT3-based GON definition.
Results: :
2516 participants were included in this study of whom 66 had GVFL in at least one eye. The HRT3 linear cup-disc ratio (LCDR) variable showed the highest sensitivity at the required specificity of 0.975 and the best agreement with GON defined using photography. The LCDR 97.5th percentile was 0.67 for small discs (up to 1.5 mm2), 0.73 for normal discs and 0.79 for large discs (above 2.0 mm2). With these cut-off values, disc-area adjusted LCDR had a sensitivity of 0.31. The HRT3 Glaucoma Probability Score and previously published linear discriminant functions showed a lower sensitivity than LCDR at this specificity
Conclusions: :
LCDR, stratified for disc area, turns out to be the most suitable candidate for an HRT3-based GON definition for epidemiological purposes in this white population.
Keywords: clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • clinical (human) or epidemiologic studies: risk factor assessment • optic disc