Evaluating Observer Variability in Tear Breakup Time for Improved Dry Eye Assessment

Abstract

Purpose : Dry Eye Disease (DED) is a prevalent condition necessitating an informative diagnostic marker. Despite longstanding use of fluorescein Tear Breakup Time (TBUT) as a marker for DED (especially evaporative) its utility remains limited by suboptimal sensitivity and specificity as well as test variability. While previous studies primarily examined variance across different TBUT trials, our study looks at variability due to observer-dependent factors.

Methods : All study activities were done in compliance with the Stony Brook IRB. Haag Streit Imaging module 910 recorded 8 TBUT videos under continuous illumination of the entire corneal tear film. A consensus score (CS) for each video was jointly determined by 2 authors (RH,TC) for Time and Zone of first TBUT (unlimited reviews of each video). 5 more videos were created by inverting or accelerating select recordings. Non-Invasive Keratographic Breakup Time (NIKBUT) was also obtained for objective comparison.

Survey videos were presented in random order to participants on a single monitor in standardized illumination settings. Participants viewed the 13 videos rating time and zone of first TBUT for each. Participants only had 1 view of each video. Results were analyzed in Excel for Correctness (COR) - agreement to CS; and Consistency (CON) - agreement identifying same time and zone in inverted/modified videos.

Results : Participants included: 10 ophthalmology attendings, 13 ophthalmology residents, and 3 medical students at 3 institutions. Results shown in Table 1. COR scores were similar for each Group showed and good agreement with CS. No clear pattern for Time or Zone scoring between CS and participant scores was seen. Participants exhibited both greater COR and CON rating the Zone of breakup rather than the Time (p < .001). Attendings had statistically better COR and CON for both Time and Zone compared to trainees. (p = .048). NIKBUT detected the first breakup significantly earlier than TBUT (p = .007).

Conclusions : TBUT measures between different examiners show good correlation validating its use. COR scores improved with level of training suggesting that teaching programs may improve TBUT assessments. The significant temporal disparity between NIKBUT and TBUT may suggest a higher sensitivity of automated measures, or suggest the techniques measure different aspects of tear film function. The findings highlight opportunities for refinement in DED evaluation.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.