Abstract
Purpose :
The IOLMaster (Carl Zeiss, CA) is often used for tracking changes in axial length (AL) in myopic progression. The number of measurements for AL values reported in the literature varies from 3 to 20. We performed a prospective analysis to determine the fewest number of measurements to obtain a measurement error (ME) ≤ 0.04mm and 95% Confidence Interval (CI) ≤ 0.04mm, or approximately 0.125 Diopters (D).
Methods :
The study was conducted at 3 sites with IOLMaster software 5.0 or later. 7 subjects (5 female/2 male) were included (power =0.81). All subjects were ≤ age 40 and in good health. ME was determined by: ME=(Repeatability (RE)2 + Reproducibility (RP)2)½ : RE=intraobserver variability, RP=intersession variability. Standard deviation (SD) of one observer measuring the same subject 10 times with 5,10,15, 20 measurements was used for RE. SD of the difference between session 1 and 2 was used for RP. Composite AL values were recorded. No individual readings were eliminated or manually adjusted, per instructions manual. Two-sample, two-tailed t-tests were used to determine statistical significance (p ≤ 0.05). 95% CIs were calculated for the difference between sesssions, and compared to the pre-determined criteria. 1mm in AL was set equal to 3D.
Results :
Intrasession 5,10,15, 20 measurements AL (mm) were 23.14±0.030; 23.14±0.029; 23.12±0.022; 23.13±0.016 (all p>0.05). Intersession 5,10,15, 20 differences in AL (mm) were -0.028±0.039; 0.030±0.038; -0.042±0.051; -0.0007±0.008. Only the 20 measurement was statistically significantly different from all the others (p<0.008 after Bonferroni correction). For 5,10,15,20: ME (mm) 0.050, 0.048, 0.056, 0.018; ME (D) 0.149, 0.143, 0.167, 0.054. The 95% CIs for the difference between sessions were: 5: -0.045, -0.10: Spread = 0.035mm = 0.103D; 10: 0.014, 0.046: Spread = 0.032mm = 0.095D; 15: -0.063, -0.020: Spread = 0.043mm = 0.129D; 20: -0.004, 0.005: Spread = 0.009mm = 0.03D.
Conclusions :
Only 20 measurements met the ME criteria. For difference between sessions, 20 measurements was better both in terms of statistically different from others and having a narrower 95% CI. Only 20 measurements met both criteria. When using the IOLMaster with software 5.0 or greater to track changes in AL over time for myopic progression, the most appropriate number of measurements to take is 20 per session to have 95% confidence that the true measurement is within 0.04mm, or 0.125D.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.