Optical coherence tomography reader agreement in neovascular age-related macular degeneration.

Purpose: To determine the interreader and intrareader agreement at the Optical Coherence Tomography (OCT) Reading Center at Duke for images produced for an interventional neovascular age-related macular degeneration (AMD) clinical trial.

Design: Retrospective, observational case series.

Methods: OCT was performed using the Stratus OCT Fast Macular Thickness Map (Carl Zeiss Meditec, Dublin, California, USA) scan mode and a 7-mm line scan centered on the fovea. Experienced OCT readers, certified to grade scans according to a standardized protocol, independently determined whether a scan was gradable. If gradable, each of the scans was graded for multiple morphologic characteristics. In addition, retinal thickness, subretinal fluid thickness, and choroidal neovascularization (CNV) thickness at the fovea were measured for each gradable scan. Interreader agreement was determined among three reader pairs. Readers regraded a subset of scans and intrareader agreement was determined.

Results: The interreader agreement was high for scan gradability and scan grades among three reader pairs, ranging from 84% to 100% and from 84% to 96%, respectively. Similarly, the intrareader agreement for scan gradability and scan grade, including comparison of adjudicated scan grades with readjudicated scan grades, was high and ranged from 91% to 100% and from 79% to 98%, respectively, except for scan grades of retinal pigment epithelium atrophy, when CNV was present. Highly reproducible results also were found for quantitative thickness measurements.

Conclusions: Well-trained OCT readers can grade independently, with a high level of interreader and intrareader agreement, multiple morphologic parameters of OCT scans obtained from eyes with neovascular AMD. Reproducible Stratus OCT scan data will be valuable to monitor treatment efficacy in interventional clinical trials of neovascular AMD.