Impact of quality indicators on variability of keratometry measurements using a swept-source OCT based optical biometer

Purpose: To characterize the variability of keratometry measurements on the IOLMaster 700, and relate it to device image quality indicators (QI). Setting: Two academic centers and one private practice. Design: Multicenter, retrospective consecutive case series. Methods: Measurements from three sites, obtained between December, 2015 and July, 2023 were included. Surgery-naïve phakic eyes with same-day sequential measurements on the same eye were identified. Repeat measurement pairs were grouped by IOLMaster QIs (success vs. warning), and changes in mean standard (∆Kmean) and total (∆TKmean) keratometry as well as standard (∆Kastig) and total (∆TKastig) astigmatism vectors were calculated. Results: Analysis was performed on 3,222 eyes of 1,890 patients. Measurement 'success' was associated with a smaller ΔKmean (0.09 ± 0.14 D) and ΔTKmean (0.11 ± 0.16 D) when compared to pairs in which both measurements had a 'warning' [0.25 ± 0.32 D and 0.14 ± 0.17 D, respectively; (p < 0.0001)]. A similarly smaller ∆Kastig (0.26 ± 0.28 D) and ∆TKastig (0.28 ± 0.30 D) was observed with measurement 'success' versus 'warning' [0.77 ± 0.79 D and 0.42 ± 0.41 D, respectively (p < 0.0001)]. Even when both measurements were successful, the proportion of measurement pairs that had a ∆Kastig > 0.50 D increased from 14% to 24% to 32% when Kmean standard deviation (SD) was ≥ 0.01, 0.05, and 0.10 D, respectively. Conclusions: When measurement quality is poor, total keratometry varies less than standard keratometry measurements. Clinicians may use the SD of Kmean/TKmean to estimate the repeatability of measurements and balance this against their tolerance for performing repeat measurements.