Between Eye Correlation Of Ocular Parameters

In ophthalmic statistics, it is accepted that ocular parameters of any given eye are more correlated with the fellow eye than with an eye of another person. This poses a challenge in ophthalmic research. There is relative scarcity in data concerning the extent of correlation in different ocular parameters. The aim of this study is to quantify the correlation of different ocular parameters between the eyes of the same individual.

Ophthalmology department in one academic tertiary medical center (Shamir Medical Center).

This is a historical cohort study. We extracted data from OCT examinations (Heidelberg -Spectralis OCT ;Heidelberg Engineering GmbH, Dossenheim) and biometry (IOL Master700 ;Carl Zeiss Meditec AG, Germany) of all patients examined in a ten-year period (2011-2021). This cohort of patients include a diverse population of both healthy individuals and those affected by ocular disorders, in order to better represent the typical population studied in ophthalmologic research. A single exam (two eyes) of each patient was included taken at the same time. Pearson’s r values were calculated to estimate the extent of correlation.

A total of 17,212 patient were included. Mean age was 73.12.7 years. [I.H1] All examined parameters were highly statistically significantly correlated between eyes. Low correlation was seen in central macular thickness (n=17,212, 304.92±101 vs 305.68±103 µm, r=0.189, p< 0.1*10^-36), and a small correlation was seen in anterior chamber depth (n=8,773, 3.25±0.6 vs 3.23±0.6 mm, r=0.379, p< 0.1*10^-36) and retinal nerve fiber layer thickness (n=8,244, 92.6±29 vs 92.0±30 µm, r=0.479, p< 0.1*10^-36). High correlation was seen in central corneal thickness (n=9,231, 530.1±0.4 vs 531.0±0.5 µm, r=0.754, p< 0.1*10^-36), and very high correlation was seen in axial length (n=8,567, 23.67±1.5 vs 23.64±1.5 mm, r=0.900, p< 0.1*10^-36).

In a retrospective analysis of OCT and biometry measurements of over 10,000 patients over a ten-year period, all examined parameters were positively and highly statistically significantly correlated. The magnitude of correlation ranged from low (0.2) to high (0.9) in different parameters. It likely that using both eyes in an outcome analysis will cause a consistent underestimation of variance and introduce bias. We therefore recommend adjusting for inter-eye correlation in all ocular outcome measures.