Diagnostic accuracy and reproducibility of tendency oriented perimetry in glaucoma

Abstract

Purpose: To evaluate the diagnostic capability of tendency oriented perimetry (TOP) in glaucoma.

Methods: A): The diagnostic accuracy of mean defect (MD), square-root of the loss variance (s LV), and number of pathologic points (NPP) was calculated in 295 normal and 414 glaucoma eyes (179 early, 112 moderate, and 123 advanced) examined with TOP. B): Threshold fluctuation (F) and its relationship with the loss variance (LV) was measured in 34 normal and 33 glaucoma eyes (mean MD=3 dB; SD=3.9) for TOP and for full-threshold perimetry (FT). C): Twenty-eight eyes with stable glaucoma (mean MD=9.5 dB; SD=7.2) were examined six times to quantify LV error. D): TOP and FT were tested with the simulation program PeriSim using different behavior models.

Results: A): The best diagnostic index in early glaucoma (MD<6dB) was sLV (specificity=90.2%, sensitivity=84.9). The three indices had similar precision in moderate and severe glaucoma. B): Threshold fluctuation and sLV were better correlated in TOP (r=0.72, p<0.01) than in FT (r=0.62, p<0.01). For normal subjects, in FT the incidence of F<2 dB was 8.82% and s LV<1.5 dB 5.88%. The same frequencies in TOP were 67.65% and 55.88%. C): Averaging six examinations reduced the sLV val ue by 22%. D): The threshold estimation error increased 1 dB in TOP in relation to FT for the same patient's behavior, but the error in TOP was lower than i n FT when the worst behavior was modeled.

Conclusions: TOP is a good discriminator between glaucoma and normality. Perimetry results overestimate the real sLV value. TOP's high diagnostic ability is probably associated to the algorithm design and to less contaminating influences during the examination.