Purchase this article with an account.
Jayme A. Vianna, Alexandre S. Reis, Lucas P. Vicente, Marcelo Hatanaka, Paul H. Artes; Assessment of Glaucomatous Optic Disc Damage by Ophthalmology Residents at the University of Sao Paulo. Invest. Ophthalmol. Vis. Sci. 2012;53(14):644.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To examine differences in performance in diagnosing glaucomatous optic disc damage among ophthalmology residents in years 1 to 3.
In this cross-sectional study, 38 residents (12 in the first, 14 in the second and 12 in the third year of training) tested themselves with the Discus software (Denniss, Opt Vis Sci, 2011). This software displayed 100 non-stereo optic disc photographs from patients with manifest glaucoma (altered visual fields, n=20) and suspected disease or ocular hypertension (normal visual fields, n=80), each for up to 10 seconds. The observers rated the likelihood of optic disc damage on a 5-point scale (definitely healthy [-2], probably healthy, not sure, probably damaged, definitely damaged [+2]). Diagnostic performance was evaluated with reference to visual field loss (Area under the ROC - AuROC), and as the rank correlation of responses to those from an expert panel. Criterion (the likelihood to call damage) was evaluated from the mean of the responses.
Diagnostic performance, response latency, decision criteria for individual participants, and summary statistics for each year, are shown in Figure. Median AuROC tended to be smaller in first (0.69) than second and third years (0.74 for both, p=0.61), and median response latency larger in the second (7.6 s) than in first (5.6 s) and third years (5.4 s, p=0.04). Rank correlation with the expert panel (0.66, 0.65 and 0.65 in first, second and third years, respectively, p=0.86), and decision criteria (1.92, 1.85 and 1.81, in first, second and third years, respectively, P=0.52) were similar among the 3 years. There was a moderate correlation between the two measures of performance (AuROC and correlation with experts, rho=0.61, p<0.001), but no relationship between either performance measure and decision criteria (rho=-0.13 and -0.01, p>0.10).
Despite considerable heterogeneity in performance among residents in the same year, residents in the second and third years of training tended to perform better than those in the first. Discus provides a simple, rapid, and objective assessment of performance that should be useful in many training programs.
This PDF is available to Subscribers Only