Introduction
Interexpert Agreement of Plus Disease Diagnosis in Retinopathy of PrematurityChiang MF, Jiang L, Gelman R, Du YA, Flynn JT
Arch Ophthalmol. 2007;125:875-880
Summary
Twenty-two experts (17 pediatric ophthalmologists and 5 retinal surgeons) on the management of ROP were asked to interpret a set of 34 wide-angle retinal photographs from infants with ROP. These experts were asked to rate the photographs in terms of diagnosis (plus, pre-plus, neither, or cannot determine) and quality (adequate, possibly adequate, or inadequate for diagnosis). Absolute agreement among participants for each image was determined according to a 3-level (plus, pre-plus, or neither) or 2-level (plus or not-plus) categorization.
All participants reviewed all 34 images, for a total of 748 responses. There were 18 "cannot determine" responses (2%) and 20 responses (3%) deeming the image quality inadequate for diagnosis. The 3-level diagnosis was made by at least 90% of the experts for 6 images (18%) and by at least 80% of the experts for 7 images (21%); the 2-level diagnosis was made by 90% of the experts for 20 images (59%) and by at least 80% for 24 images (71%).
Comment
This study demonstrates the difficulty of determining the presence of plus disease on the basis of current criteria. Unfortunately, one of the parameters for treatment of ROP is based on the presence or absence of plus disease. It should be noted, however, that this study did not mimic true clinical scenarios, in which a patient suspicious for plus disease but not conclusively having that feature may be followed and examined serially. What the study may end up demonstrating most effectively are both the limitations of screening patients for ROP severity with fundus photography and the difficulty of developing a computer- or image-based algorithm to diagnose these cases.
Abstract
Supported by an independent educational grant from Genentech