Diagnostic performance of modern imaging instruments in glaucoma screening

To evaluate the applicability of imaging devices (spectral-domain optical coherence tomography (Cirrus SD-OCT), scanning laser polarimetry (GDx) and scanning laser ophthalmoscopy (Heidelberg Retinal Tomograph, HRT3)) for glaucoma screening in a middle-aged unselected population.

Participants of the population-based Northern Finland Birth Cohort Eye Study, aged 45 to 49 years, underwent a comprehensive eye examination including modern imaging with five methods (retinal nerve fibre layer (RNFL) and macular ganglion cell layer +inner plexiform layer (GCIPL) analysis and their combination with SD-OCT, GDx and HRT). The performance of the automated classification of the imaging devices was assessed using a clinical glaucoma diagnosis as reference, that is, the '2 out of 3' rule based on the evaluation of optic nerve head and RNFL photographs and visual fields.

We examined 6060 eyes of 3039 subjects; in the clinical evaluation, glaucomatous damage was found in 33 subjects (1.1%) in 43 eyes. The following sensitivities were obtained; RNFL analysis (53%), GCIPL analysis (50%), OCT combination analysis (61%), GDx (56%) and HRT (31%) with corresponding specificities of 95%, 92%, 90%, 88% and 96%. The area under the curve values were 0.76, 0.73, 0.75, 0.75 and 0.73, respectively. Post-test probabilities of glaucoma after positive imaging finding with each of these methods in this unselected population were 11%, 7%, 6%, 5% and 7%, respectively.

Screening capabilities of the OCT, GDx and HRT were rather similar. The accuracy of all evaluated parameters was only moderate and thus screening with these parameters alone is not reliable.