Comparing the detection and agreement of parapapillary atrophy progression using digital optic disk photographs and alternation flicker

Parapapillary atrophy (PPA) progression has been associated with progressive glaucoma, but has proven to be difficult to assess clinically. We compared inter- and intra-observer agreement using a novel automated alternation flicker technology and side-by-side digital photography inspection for the evaluation of PPA progression.

Consecutive patients with serial digital optic nerve photographs at least 1 year apart were included. Two graders (NR, BV) masked to image chronology assessed a set of photographs for progressive PPA using predefined criteria based on reference photographs containing mild, moderate, extensive or no PPA progression. At a separate session, the graders evaluated photographs using alternation flicker (EyeIC, Narberth, PA, USA) applying the same criteria. The order of patients and technique was randomized. Graders then assessed the same set of flickers and photographs a second time with the order of presentation reversed. The main outcome measure was the assessment of progressive PPA as identified by alternation flicker and digital photography inspection. Inter- and intra-observer agreement using each technique was assessed using the kappa statistic. A bootstrap method for comparing correlated kappa coefficients was used to assess statistical significance.

Serial photographs from 131 eyes of 68 patients were evaluated. Both graders identified significantly more cases of PPA progression using flicker compared to photography (27–34% vs 8–13%; both p ≤ 0.003). Inter-observer agreement using flicker was better than using photographs (κ = 0.52 vs 0.18, p = 0.02). Intra-observer agreement was similar for both graders (photos: κ = 0.58 vs 0.57, p = 0.97; flicker: κ = 0.61 vs 0.70, p = 0.37). When progression was assessed by the number of progressive quadrants identified by each grader using a weighted kappa statistic, flicker inter-observer agreement was still moderate (κ = 0.45) and significantly better (p = 0.01) than photography, which showed poor agreement (κ = 0.15). Intra-observer agreement with a weighted kappa for quadrant progression was also similar for both graders (photos: grader 1 κ = 0.53 vs grader 2 κ = 0.52, p = 0.92; flickers: grader 1 κ = 0.58 vs grader 2 κ = 0.69, p = 0.22).

Flicker identified more cases of progressive PPA than photographic review. Agreement between observers was significantly higher when using the automated flicker technology.