The Impact of the Perimetric Measurement Scale, Sample Composition, and Statistical Method on the Structure-function Relationship in Glaucoma

PurposeTo investigate the impact of perimetric measurement scales, sample composition, and 2 statistical methods on the structure-function relationship in glaucoma. MethodsThree hundred eighty-five eyes of 385 participants, covering the full spectrum of the disease from healthy to advanced glaucoma, were tested with the Heidelberg Retina Tomograph and standard automated perimetry (Swedish Interactive Thresholding Algorithm) within a 6-month period. The associations (R2) between neuroretinal rim area and perimetric data expressed in logarithmic (dB) and linear (1/Lambert) units were measured for linear and curvilinear (quadratic and logarithmic) models. The best-fit model resulting from 2 statistical analyses (χ2 coefficient and analysis of the residuals) were compared in the visual field areas corresponding to the 6 Heidelberg Retina Tomograph sectors and in the central visual field. These analyses were performed on the complete sample and on a subset of the study population composed only of healthy controls and patients with definite glaucoma. ResultsThe R2 values ranged from 0.01 to 0.32 with the strongest associations in the supero-temporal and infero-temporal optic disc regions. When perimetric data were expressed in dB units, the curvilinear models generally explained more variance than the linear model. When perimetric data were expressed in 1/Lambert units, differences were observed between the 2 statistical and subgroup analyses. ConclusionsThe perimetric measurement scale is one of many factors influencing the structure-function relationship. It is also influenced by the composition of the sample and by the statistical method of analysis. Expressing perimetric data in linear units does not always translate into linear structure-function relationships.