Repeatability and Reproducibility of the Dry Eye Analyzer in Participants With and Without Symptoms of Dry Eye.

PURPOSE: Reliable evaluation and consistency between examiners and sessions are essential for dry eye diagnosis. This study examined the inter-examiner reproducibility (IER), inter-session repeatability (ISR) and within-subject variability (WSV) of the Dry Eye Analyzer (DEA) in participants with and without dry eye symptoms. METHODS: Participants with Dry Eye Questionnaire-5 (DEQ-5) scores ≥6 were classified as symptomatic. Non-invasive tear breakup time (NITBUT), Meibomian gland (MG) loss, tear meniscus height (TMH), interferometry and conjunctival redness were measured by two examiners on the same day for IER. Examiner 1 repeated measurements 1-2 weeks later for ISR. Analyses included Wilcoxon tests, Spearman correlation, weighted kappa (κw), within-subject standard deviation (Sw), repeatability limits (2.77Sw), intraclass correlation coefficients (ICCs) and Bland-Altman analysis. RESULTS: IER included 84 participants (mean age 23 ± 2 years, range 18-34 years; 42 symptomatic). Examiner 1/Examiner 2 values were 7.2 ± 3.2/8.9 ± 3.0 s for NITBUT, 20.4 ± 7.4/20.4 ± 7.0% for upper MG loss, 27.2 ± 14.0/26.6 ± 14.0% for lower MG loss and 0.2 ± 0.0/0.2 ± 0.1 mm for TMH. Outcomes were positively correlated without significant differences for most comparisons; NITBUT differed significantly overall and in symptomatic participants. Agreement ranged from fair to good (κw 0.30-0.63), ICCs from 0.51 to 0.99 and Sw was <3.40. Mean NITBUT bias was -1.68 s overall and -2.90 s in symptomatic participants; 93-98% of observations fell within the limits of agreement. ISR included 68 participants (mean age 23 ± 2 years, range 18-32; 36 symptomatic). Session 1/Session 2 values were 7.1 ± 3.2/6.8 ± 3.1 s for NITBUT, 20.7 ± 7.4/20.7 ± 8.1% for upper MG loss, 27.0 ± 14.6/26.9 ± 13.9% for lower MG loss and 0.2 ± 0.0/0.2 ± 0.0 mm for TMH. Outcomes were positively correlated without significant differences. Agreement ranged from fair to very good (κw 0.28-1.00), with Sw < 3.14. CONCLUSIONS: DEA reliability was parameter-dependent. MG-loss measurements showed excellent IER and low WSV, whereas NITBUT showed moderate reliability and examiner-related bias. NITBUT and TMH should be interpreted cautiously during follow-up.