[The 'cup-to-disc ratio': a comparison of TopSS, HRT II and subjective findings]

One of the major shortcomings of longitudinal studies is the fact that instruments and technologies which were used at the beginning of the sampling might be replaced by others during the course of the study. The aim of the present work is to evaluate the relationship of optic disk parameters assessed with different methods and to test statistical possibilities of transforming these parameters into each other.

The 'cup-to-disc ratio', the 'total disc area' and the 'neuroretinal rim area' of the human optic nerve of 131 eyes of 66 patients participating in the Salzburg-Moorfields Collaborative Glaucoma Study were assessed with two different laser scanning methods (TopSS and HRT II. The 'cup-to-disc ratio' was also determined subjectively by ophthalmologists of the glaucoma department. To compare the three data sets, the method of Bland-Altman, paired t-tests, as well as regression analyses were applied.

The 'cup-to-disc ratios' were: HRT II(R): 0.26 (95 % CI: 0.23 - 0.28), subjective assessment of 'cup-to-disc ratio': 0.33 (0.30 - 0.36) and TopSS: 0.43 (0.40 - 0.46). All three results are statistically significantly different (pairwise comparisons, p < 0.00001 each). Bland-Altman analysis shows that the differences of both objective methods exceed a magnitude that can be accepted for clinical purposes. The regression analyses reveal that the slope parameters are significantly different from 1.0. No regression models could be found with sufficiently small differences.

There are significant differences in the 'cup-to-disc Ratios' observed between the three methods. Therefore the application of a 'correction factor' cannot be advocated. Although such factors allow the means of the samples to agree, the differences of individual measurements still remain too large, to be useful for practical purposes.