Repeatability of tear film lipid layer interferometry measurements: A randomized, crossover study

This study aimed to assess intrasession repeatability, diurnal intravisit and day-to-day intervisit reproducibility of qualitative tear film lipid layer interferometry grading obtained from the Keratograph 5M (Oculus Optikgeräte GmbH, Wetzlar, Germany), and automated quantitative layer thickness measurements obtained from the TearScience LipiView II Ocular Surface Interferometer (Johnson & Johnson Vision, Milpitas, CA), in community residents, not stratified by dry eye or blepharitis status.

Forty community residents (24 women, 16 men; age mean ± standard deviation, 36 ± 14 years) were recruited in an investigator-masked, randomized, crossover study. Participants attended two sessions 6 hours apart on the same day and a third session on a separate day. During each session, tear film lipid layer grading and thickness measurements were performed in triplicate, in a masked manner, with 10-minute intervals between each measurement.

There were no significant differences in lipid layer grading and thickness measurements in all pairwise intrasession, diurnal intravisit, and day-to-day intervisit comparisons (all p≥0.19), and positive correlations were observed in all cases (all p<0.001). The weighted Cohen κ coefficients for intrasession repeatability and diurnal intravisit and day-to-day intervisit reproducibility of qualitative lipid layer grading ranged from 0.670 to 0.867. The test-retest repeatability of quantitative lipid layer thickness measurements ranged from 4.5 to 11.7 nm, and the Bland-Altman biases were within the order of 3 nm in all pairwise comparisons. The intraclass correlation coefficients for lipid layer thickness measurements exceeded 0.80 for all intrasession comparisons and were between 0.60 and 0.80 for all diurnal intravisit and day-to-day intervisit comparisons.

Both qualitative grading and automated quantitative thickness measurements of tear film lipid layer interferometry demonstrated relatively favorable levels of intrasession repeatability and diurnal intravisit and day-to-day intervisit reproducibility. The study findings would support the use of either qualitative or quantitative measurement in clinical and research settings, as a reliable tool for assessing lipid layer interferometry.