Logistic regression analysis in the combination of diagnostic methods in glaucoma

Author(s)

Leonardo Reichmann Fasolo ^[1] ; Luiz Alberto Soares Melo Júnior ^[2] ; Aline Silva Barbosa ^[3] ; Rodrigo Antônio Brant Fernandes ^[4] ; João Antonio Prata Junior ^[5]

Affiliation

^[1] Universidade Federal de São Paulo. Escola Paulista de Medicina. Departamento de Oftalmologia - Brasil

^[2] Universidade Federal de São Paulo. Escola Paulista de Medicina. Departamento de Oftalmologia - Brasil

^[3] Universidade Federal de São Paulo - Brasil

^[4] Universidade Federal de São Paulo. Escola Paulista de Medicina. Departamento de Oftalmologia - Brasil

^[5] Universidade Federal do Triângulo Mineiro - Brasil

• Document type: Journal article
• Source: Arquivos Brasileiros de Oftalmologia; v. 76, n. 6, p. 341-344, 2013-12-00.

Abstract

PURPOSES: In this research was studied the diagnostic ability of the Heidelberg Retinal Tomograph (HRT II), GDx nerve fiber analyzer (GDx), Short-wavelenght Automated Perimetry (SWAP) and Frequency-doubling technology (FDT), all of them together and also separately in the diagnosis of glaucoma. METHODS: Sixty glaucomatous patients and 60 normal ones were submitted to examination with the technology of HRT II, GDx, FDT and SWAP. In the interpretation of HRT, based on the Moorfields regression analysis, were considered abnormal when at least one area of the neuroretinal ring was outside the bounds of normality. GDx was defined as altered, when it was observed that at least one index was considered by equipment program as outside of normal limits, excluding the index symmetry, or when "the deviation from normal graph" appeared a quadrant with significance below 5%. The FDT was considered abnormal when at least one region tested presented with severe defects, or two contiguous moderate defects in at least two reliable tests. For SWAP was adopted the abnormality criteria proposed by Anderson. Logistic regression analysis was performed. RESULTS: When the technologies were studied separately, the Logistic regression analysis showed better odds ratio rates to test positive for glaucoma with HRT (22.49), followed by EAP (21.71), FDT (3.97) and GDx (2.73). When positive tests were associated with different technologies, the odds ratios increased. In cases with HRT, FDT and PAA tests, outside normal bounds, the odds ratio was 252.6. Concerning the HRT, GDx and PAA the result have changed to 173.1. When we associate positive tests from different technologies, the odds ratio of glaucomatous patients have increased considerably, reaching 689.7 with all the tests outside normal bounds, which have happened with 26 patients of this study. CONCLUSIONS: The Logistic Regression analysis confirmed that the abnormal tests of HRT and PAA have shown to be the highest reasons for glaucoma occurrence. The association of abnormal tests increased the odds ratio, especially when the HRT and PAA were out of normal bounds. (AU)