ObjectiveTo explore participant-level biological attributes and scan-level methodological attributes associated with retinal nerve fiber layer (RNFL) thickness variability in a population-based sample of elderly United States adults. DesignCross-sectional analysis using data from the Framingham Heart Study. Participants1,347 eyes from 825 participants with at least one optical coherence tomography (OCT) scan and axial length data were included. MethodsThree or more successive RNFL scans of each eye of each participant were obtained in a single session. Multivariable linear mixed models were employed to explore the associations between average RNFL thickness with participant-level biological attributes (age, gender, race, ethnicity, and axial length) and scan-level attributes (signal strength [SS]) as independent variables in the whole population as well as a sub-sample of adults with no self-reported history of glaucoma. Similar analyses were designed to assess methodological variability with average within-eye standard deviation (SD) for repeated scans as the dependent variable. Main Outcomes1) Biological variability: average RNFL thickness, 2) Methodological variability: average within- participant SD across repeated scans. ResultsAge (β= −0.19 microns/year, [95% CI: −0.29, −0.09]), female gender (β= +1.48 microns vs. male, [95% CI: 0.09,2.86]), axial length (β= −1.24 microns/mm of greater length, [95% CI: −1.80, −0.67]), and SS (β= +1.62 microns/1 unit greater SS, [95% CI: 1.16,2.09]) were significantly associated with RNFL thickness, while race and ethnicity were not (p>0.05). In analyses designed to assess methodological variability, higher RNFL thickness (β= +0.02 per micron increase, [95% CI: 0.01,0.03]), and lower SS (β= +0.19 per 1 unit lower SS, [95% CI: 0.10,0.27]) were significantly associated with greater RNFL variability. In adults with no self-reported history of glaucoma (n of eyes = 1165, n of participants = 712), female gender was not associated with RNFL, while African-American race was associated with thicker RNFL (β= +4.65 microns vs. Whites, [95% CI: 1.28,8.03]). ConclusionsRNFL thickness is lower with older age, male gender, greater axial length, lower signal strength, and Whites (as compared to African-Americans) without self-reported glaucoma. Measurement variability (SD) is higher with greater RNFL thickness and lower SS. Understanding these biological and methodological variations is important to aid in OCT interpretation.
Read full abstract