Abstract
Abstract: :
Purpose: To examine the effect of measurement variability on the agreement between visual field progression and Heidelberg Retina Tomograph (HRT) rim area (RA) progression. Methods: 198 ocular hypertensive (OHT) subjects (median age 60.2 yrs, range 32.0–78.9; median follow–up 6.1 yrs, range 2.3–7.2) and 21 normal controls (median age 65.0 yrs, range 41.3–77.4; median follow–up 5.3 yrs, range 3.1–6.8) were selected from a group of patients followed prospectively for progression. Each subject had Humphrey visual field testing and HRT imaging on > 5 occasions; 1 eye per subject was analyzed. HRT images were analyzed using a 320µm reference plane. Linear regression of global RA over time was performed. Variability was estimated from the residual standard deviation (RSD). Progression was defined as a significant negative slope >1% of baseline RA/year (p<0.01, low variability; p<0.002, high variability). Visual fields were analyzed by pointwise linear regression of sensitivity over time. Progression was defined by the 3–omitting criterion (Gardiner SK – IOVS 2002). Variability was estimated using the RSD of each test point, averaged across the entire field. As variability increases as visual field sensitivity decreases, the RSD values were normalized using the equation provided by Henson (IOVS 2000) loge(SD) =–0.081*Sensitivity+3.27. Subjects were classified as high (H) or low (L) variability for RA and visual fields based on a 50th centile RSD cut–off. Subjects were categorized as follows: high variability RA and fields (HH), low variability RA and fields (LL), high variability RA and low variability fields (HL) and vice versa (LH). Results: HH–OHT (51 subjects): 4 progressed by both disc and field, 13 by field alone and 3 by disc alone. HL–OHT (50 subjects): 2 progressed by both disc and field, 3 by field alone and 3 by disc alone. LH–OHT (54 subjects): 3 progressed by both disc and field, 6 by disc alone and 2 by field alone. LL–OHT (43 subjects): 4 progressed by disc alone and 5 by field alone, with no agreement. 1 control progressed by disc in the HL group. A plot of ‘normalized’ RSD against mean sensitivity (100 subjects) showed a non–significant negative trend. Conclusions: Poor agreement was not explained by differences in variability between modalities. Field variability may have been underestimated by the method used; further methods to estimate variability need to be assessed.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • perimetry • optic disc