Purchase this article with an account.
Todd Warner Altenbernd, Joseph M Miller; The effect of repeated intra-visit measurements on the variability of nerve fiber layer thickness using the Spectralis spectral domain ocular coherence tomography (OCT). Invest. Ophthalmol. Vis. Sci. 2014;55(13):4734.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To determine if three measurements of nerve fiber layer thickness on each visit will reduce the 95% confidence interval by >=5 um as compared to a single measurement per visit.
46 eyes from 46 consecutive patients were evaluated for variability in the measurement of optic nerve fiber layer thickness using the Spectralis spectral domain OCT on high speed acquisition mode. Each of the 9 standard sectors were evaluated. Exclusion criteria included a signal to noise value of <15 or poor segmentation delineation anywhere within the scan. If both eyes were eligible a random numbers generated protocol was used to select one eye. Statistical analysis was performed using Stata 13.1, statistical software from StataCorp. Bootstrapped estimates of the standard error of the mean were calculated for the intra-visit variation and total (intra-visit + inter-visit) variation. Each was calculated with repeated bootstrap estimates using 50000 repetitions to give a constancy in the standard error of two to three significant figures. Intra-visit variability of the bootstrapped normalized standard errors of the mean were assessed by using strata for both the "eye" and "visit". Here a sample size of 1 represented a single measurement per visit and a size of 3 corresponded to 3 measures on each visit. Similarly, total variation was assessed by using only the "eye" as a strata. Here the bootstrapped random selections were taken from both visits. Sizes of 2 and 6 represented 1 and 3 measurements per visit respectively.
Increasing the number of measurements per visit from one to three decreased the standard error of the mean for the Intra-visit and total standard errors by ~73% in each of the 9 sectors, Table 1. The intra-visit standard deviation is of the same order of magnitude as the total standard deviation with the Intra-visit : Total variation ranging from 0.64-0.85 : 1, Table 1. The 95% confidence interval decreased by 3.57-6.27 um in the 9 sectors with 6 of the 9 being >= 5 um, Table 2.
Taking 3 measurements per visit verses 1 reduced the confidence interval to a clinically significant degree in most sectors and may allow detection of change at an earlier time as well as increase the number of interval changes possible between the current nerve fiber layer thickness and the floor thickness.
This PDF is available to Subscribers Only