Physiology of cumulative defect curves; consequences in glaucoma perimetry

Purely diffuse field loss is a controversial sign of early glaucoma. Cumulative defect curves have been proposed to distinguish between diffuse and localized visual field loss. In such curves, pointwise deviations from the normal reference field are ranked. The deviations are then compared with normative limits for each rank. The ranking leads to a complete loss of spatial information and ignoration of pointwise physiological threshold variability. We believed that this might impair the curves' intended ability to distinguish diffuse and localized types of defects. The lower 5% normative limit was estimated for each rank in the cumulative defect curve using Humphrey 30-2 tests of 88 subjects from the Statpac normal database. For each rank we identified those two test points (among the total of 6512 (74 test points in each of 88 subjects)) from which the 5% limit was calculated. The average eccentricity, eccrank, of those two points was then determined at each rank. For lower to intermediate ranks eccrank was spread out from center to periphery. When instead the 20 highest ranks (most depressed points) were considered eccrank was much concentrated peripherally. Our normative limits were subsequently used to evaluate field results in 101 patients with glaucoma. In this glaucoma evaluation group, the largest deviations were often encountered in central or paracentral scotomas, in sharp contrast to the more peripheral position in the normative group. These deviations often failed to reach the cumulative defect curve limits which originated from points in the mid-periphery where normal variability is large. At present, cumulative defect curves are not optimized to distinguish diffuse and localized field loss.