Idiosyncratic Number Use is a Source of Inter-observer Variability in Glaucomatous Optic Disc Assessment

Abstract

Abstract: : Purpose: Idiosyncratic number use is a well-documented psychophysical phenomenon that describes some observers' inconsistency in estimating the relative magnitudes of a range of stimuli (ie, nonlinear scaling). This study investigates whether idiosyncratic number use in cup/disc (C/D) ratio assessment accounts for some of the inter-observer variability associated with this task. Methods: 21 ophthalmologists viewed and assigned C/D ratios to a series of optic disc photographs spanning the spectrum from C/D 0.1 to 0.9. They also viewed a series of straight lines of various lengths, and assigned to each line a number that corresponded to their impression of the line's magnitude. From the line length data, we determined each observer's deviation from perfect judgment of relative magnitudes (ie, idiosyncratic number use). We then assessed the observers' accuracy in judging C/D ratios before and after mathematically correcting judgments based on each observer's extent of idiosyncratic number use. Results: Before correction for idiosyncratic number use, the slope of the regression function arising by plotting judged C/D ratio against true C/D ratio was 0.84 (perfect agreement would give a slope of 1.0). After correction for idiosyncratic number use, the slope improved to 0.90 (p<0.05 by paired t-test). Conclusions: Idiosyncratic number use accounts for some of the inter-observer variability in the clinical assessment of C/D ratio.

Keywords: optic disc • clinical (human) or epidemiologic studies: sys • imaging/image analysis: clinical