Variability in Spectral-Domain Optical Coherence Tomography over 4 Weeks by Age.

Kleins Lab // Publications // Ronald Danis // Jun 01 2016

PubMed ID: 27128499

Author(s): Howard KP, Chandler CS, Huang Y, Klein BE, Lee KE, Meuer SM, Danis RP, Danforth LG, Gangnon RE, Klein R. Variability in spectral-domain optical coherence tomography over 4 weeks by age. Ophthalmic Epidemiol. 2016 Jun;23(3):193-201. doi: 10.3109/09286586.2015.1119285. Epub 2016 Apr 29. PMID 27128499

Journal: Ophthalmic Epidemiology, Volume 23, Issue 3, 06 2016

PURPOSE To quantify variation in spectral-domain optical coherence tomography (SD-OCT) measures of total retinal thickness (top of inner limiting membrane to top of retinal pigment epithelium, RPE) and RPE thickness measures over a 4-week period and by age.

METHODS A total of 76 volunteers aged 40-85 years were seen at three visits over 4 weeks. Two Topcon SD-OCT scans were taken at each visit. Following grid re-centration, total retinal and RPE thickness were determined in nine subfields. Multilevel modeling was used to quantify variance between scans and by age.

RESULTS In the central circle, mean total retinal thickness was 237.9 µm (standard deviation, SD, 23.5 µm) and RPE thickness was 46.0 µm (SD 5.3 µm). Intraclass correlation coefficient in the central circle was 0.988 for total retinal thickness and 0.714 for RPE thickness. Pairwise measures taken within 4 weeks were strongly correlated (p > 0.95). Within-subject variation of total retinal thickness increased significantly with age. Subjects in the oldest age group had significantly increased among- and within-subject variability in measures of RPE thickness.

CONCLUSIONS Correlation between retinal thickness measures was very high (>0.95) over a period of 4 weeks with small changes likely due to variation in measurement. Increasing variability in total retinal and RPE thickness measures with age suggest that the use of more and/or higher quality images to calculate mean thickness to reduce variability may benefit the study of these measures in older persons. This may also impact sample size calculations for future studies of SD-OCT measures in older adults.