Purchase this article with an account.
Camilla Sandberg Melin, Zhaohua Yu, Albert Alm, Eva Nuija, Per G Soderberg; Variation in Measuring Neuro-retinal Rim Area with the Heidelberg Retina Tomograph II. Invest. Ophthalmol. Vis. Sci. 2014;55(13):4763.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To estimate variation in neuro-retinal rim area (NRA) measurements determined by the Heidelberg Retina Tomograph II (HRT) for three different reference planes and to calculate the consequence of the variation on sample size required to demonstrate a significant contrast between groups.
Altogether, 24 healthy subjects were evenly divided into three groups. The NRA estimates were calculated from a three dimensional representation of the optic nerve using three strategies for reference plane estimation; standard, -320 µm,and Moorfields. NRA was estimated at 4 occasions. Totally, 5 measurements were recorded at each occasion. The variance components for subjects, occasions and measurements were estimated with an analysis of variance. Then, the sample sizes required to detect a 0.03 mm2 NRA change were estimated setting the significance level to 0.05 and the power to 0.8. Comparison of independent groups at one point in time and comparison of change over time within subject, for independent groups, were considered.
The variances for subjects were 14.3, 10.7, 13.0 x10-2 mm2 for the standard, the 320 µm, and Moorfields strategy, respectively. The variances for occasions were 2.4, 0.5, and 2.0 x10-4 mm2 for the standard, the-320 µm, and the Moorfields strategy, respectively. The variances for measurements were 6.2, 3.5 and 5.7 x10-4 mm2 for the standard, the -320µm-, and the Moorfields strategy, respectively. For comparison of independent groups at one point in time it was estimated that 314, 234 and 285 subjects per sample are required for the standard, the -320 µm-, and the Moorfields strategy, respectively. For within subject paired comparisons, for independent groups, with 3 measurements per occasion, it was estimated that 8, 3 and 7 subjects per sample are required for the standard, the -320 µm-, and the Moorfields strategy, respectively.
The variance for subjects is substantial and moderate for occasions and measurements. Means for NRA differences within subjects between occasions can be estimated with acceptable precision. Cross-sectional independent group comparisons of NRA means are not feasible, due to considerable variation among subjects.
This PDF is available to Subscribers Only