Purchase this article with an account.
Jayme R Vianna, Balwantray C Chauhan; Pilot analysis of the web-application and inter-evaluator variability for the Crowd-Sourced Glaucoma Study. Invest. Ophthalmol. Vis. Sci. 2019;60(9):6152. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The Crowd-Sourced Glaucoma Study (CSGS) is a large international collaboration that aims to identify objective visual field (VF) and optical coherence tomography (OCT) criteria that match the assessment of glaucoma likelihood by multiple glaucoma specialists. Glaucoma specialists will use a custom web-application to evaluate patient examinations and provide their assessment of the glaucoma likelihood. We conducted a pilot analysis to test the application and to obtain preliminary estimates of inter-evaluator variability.
Patients from glaucoma or general ophthalmology clinics are randomly or consecutively selected to be included in the CSGS. For each patient, 3 examinations of one eye are included: a disc photograph, a VF (program 24-2), and an OCT of the optic nerve head, peripapillary retinal nerve fibre layer thickness, and macular ganglion cell analysis (Fig. 1). Patients with non-glaucomatous pathologies or unreliable examinations are excluded. We used 30 patients initially included in the CSGS and invited current CSGS collaborators (n=47) to participate as evaluators. The application sequentially presented each patient to each evaluator in a random order, simultaneously displaying the examinations, and giving the evaluator unlimited time to review them. The evaluator then entered his/her assessment of the glaucoma likelihood, expressed continuously on a 0-100% slider scale.
Thirty-six (77%) evaluators participated of whom 32 (88%) had specialized training in glaucoma; 27 (75%) participate in the clinical care of more than 50 glaucoma patients per week, and 20 (55%) have been participating in clinical care of glaucoma patients for more than 15 years. Thirty (83%) evaluators provided feedback messages, most of which expressed satisfaction with the application. All evaluators evaluated the 30 patients, taking a median (interquartile range) of 48 (32 to 84) seconds per patient. Patients had a median age of 61 (57 to 67) years and a median VF mean deviation of -1.2 (-2.0 to 0) dB. Figure 2 shows the evaluations received by each patient. In some patients there was large variability with evaluations over the total range from 0 to 100%.
The web-application can be successfully used to obtain evaluations of glaucoma examinations. In this preliminary cohort, we observed that large inter-evaluator variability of the assessment of glaucoma likelihood can occur.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.
This PDF is available to Subscribers Only