advertisement

Editors Selection IGR 12-4

Clinical examination methods: Structure and defining glaucoma

Comment by Anja Tuulonen on:

12373 The impact of definition of primary open-angle glaucoma on the cross-sectional assessment of diagnostic validity of Heidelberg retinal tomography, Miglior S; Guareschi M; Romanazzi F et al., American Journal of Ophthalmology, 2005; 139: 878-887

Find related abstracts

Miglior et al. (517) studied how eight different criteria used to define abnormality of visual fields ('golden standard') affect the sensitivities and specificities of HRT examinations. In a literature review (1992-2003), they found that about a third of 181 papers had no IOP-based or optic nerve head definitions for POAG while VF definitions were missing in only 2 % of 181 papers. However, the definitions of POAG found in the literature included more than 30 different VF criteria which challenges the validity of this most commonly used golden standard both in diagnostic studies and every-day practice.

Comments on study design: Because of its retrospective nature, the study design does not fulfil the validity criteria for diagnostic studies and the classification of evidence does not reach high level. Although HRT is compared to an independent reference standard (visual field), the test to be studied (HRT) and reference standard (VF) were not to every patient of the clinic and it is unclear whether originally one test result influenced the decision to perform another test. In addition, the authors did not report how many patients in this sample had unreliable visual fields test and/or HRT examination and were excluded from the analysis. Further, the authors did not report how representative their patient sample was compared to the 'usual' patients at the clinic (e.g. % of all glaucoma patients visiting the clinic). Therefore, the sensitivities and specificities in unselected patient population of real life are lower and probably even more variable than reported here. Another factor that represents bias in the patient selection, is the fact that the normal subjects were younger than the glaucoma patients and suspects, and the optic disc size of the normals was statistically smaller than in the other two groups. In addition, the authors used one field only (the last field) in classifying the patients which may make their results even more variable.

Results and conclusions: Despite the weaknesses of the study design, the results very nicely show how 'everything depends on everything', i.e. the changing VF criteria led to substantial differences in sensitivities (from 0.5 - 0.8) and specificities of HRT. The message to the scientific community is clear. We need a standard definition of glaucoma and better quality diagnostic studies to make diagnostic investigations more accurate and comparable.

• ← Previous
• Next →