Reliability and interobserver variability of non‐invasive tear film parameters

Abstract

Abstract Purpose To summarize our experiences with a non‐invasive tear interference device. Methods in thirty‐seven subjects, we measured non‐invasive tear break up time (NIBUT) three times, pictured the lower tear meniscus height (TMH) and the tear film lipid layer pattern five times after complete blinks with Keeler Tearscope Plus. Three independent observers evaluated the photos. Additionally, the lipid pattern was graded by the observers with real‐time examination in thirty‐one subjects. We analysed the repeatability of the tear parameters after different blinks, and the agreement between the observers on assessments of the photos and real‐time examination. Results The results measured after different blinks did not show significant differences in case of any tear film parameters. The intraclass correlation coefficient (ICC) with 95% confidence interval of NIBUT and mean TMH were 0.605 (0.428‐0.754) and 0.901 (0.848‐0.942), respectively. There was significant difference between the measured TMH of the observers (p<0.001). The results showed significant correlations (r≥0.912, p<0.001), but the standard deviation of the three measured TMH was meanly 11.13% of the mean of them. The graded lipid patterns of observers showed significant differences (p=0.01) in case of photos, but did not (p=0.137) during real–time examination. Conclusion The Tearscope Plus is a useful tool of non‐invasive tear film examination. However, the evaluation of the parameters is partly subjective, which reduces the reliability of the methods. The real‐time examination of lipid pattern is recommended, because the kinetics of lipid layer allows additional information for grading, therefore decreasing the interobserver variability.