Sector-dependent device discordance of RNFL measurements: A prospective cross-sectional study of cross-platform...

This study aimed to characterize intra-device repeatability and sector-dependent inter-device variability of peripapillary retinal nerve fiber layer (RNFL) measurements across 5 optical coherence tomography (OCT) platforms and define the concept of sector-dependent device discordance, the non-uniform distribution of inter-device measurement bias across anatomical RNFL sectors. This prospective cross-sectional study included 38 healthy participants who underwent peripapillary RNFL imaging using 5 OCT devices: Heidelberg Spectralis, TowardPi BMIZAR 400 kHz swept-source OCT, Huvitz HOCT-1/1F, NIDEK RS-1 Glauvas, and Topcon Maestro 2. RNFL thickness was analyzed in the superior, nasal, inferior, and temporal sectors, as well as global mean thickness. Three consecutive measurements were obtained for each eye with each device to assess intra-device repeatability. Repeatability was evaluated using repeated-measures analysis of variance. Inter-device reliability was assessed using intraclass correlation coefficients, with Heidelberg Spectralis as the reference. Agreement between devices was further evaluated using Bland-Altman analysis. Excellent intra-device repeatability was documented across all platforms. However, marked inter-device variability demonstrated sector-dependent patterns: global RNFL measurements showed moderate agreement (intraclass correlation coefficient: 0.346-0.805), while nasal and temporal sectors exhibited disproportionately higher variability (absolute percentage error up to 20.76% in the nasal sector; limits of agreement width ranging from 20.23% to 35.52% in temporal). Bland-Altman analyses revealed device-specific bias signatures (Huvitz HOCT-1/1F: -8.17% global bias; NIDEK RS-1 Glauvas: +10.59%; Topcon Maestro 2: -4.47%), indicating systematic, reproducible platform-inherent differences rather than random measurement noise. While intra-device repeatability is excellent, inter-device RNFL measurements remain substantially discordant, particularly in nasal and temporal sectors, where absolute percentage error reaches up to 20.76% and limits of agreement exceed 35%, a phenomenon termed sector-dependent device discordance. Global RNFL measurements may appear concordant while sectoral measurements diverge, creating a theoretically plausible risk for false progression detection, pending validation in glaucomatous eyes. RNFL measurements are not interchangeable across platforms and are particularly unreliable for sector-specific assessment. Device-consistent monitoring is essential for glaucoma follow-up; multicenter studies and device switching scenarios warrant platform-specific normalization strategies.