Screening for glaucoma with Moorfields regression analysis and glaucoma probability score in confocal scanning laser ophthalmoscopy.

Objective: To compare the validity of Moorfields regression analysis (MRA) and glaucoma probability score (GPS) of the confocal scanning laser ophthalmoscopy (Heidelberg retina tomograph 3; HRT3) in detecting glaucomatous optic nerve damage in a screening population.

Design: Population-based, cross-sectional study.

Participants: Subjects at high risk for development of open-angle glaucoma (OAG).

Methods: All subjects underwent confocal scanning ophthalmoscopy (Heidelberg retina tomograph; HRTII) testing, visual fields testing with frequency-doubling technology perimetry (FDT), and a standard ophthalmologic examination. All HRTII images were reprocessed with HRT3. Based on an ophthalmologic examination and FDT results, eyes were classified into 4 categories: normal, possible glaucoma, probable glaucoma, and definitive glaucoma. Main outcome measures included sensitivity and specificity of HRTII/MRA, HRT3/MRA, HRT3/GPS, and combination HRT3/MRA/GPS.

Results: The left eyes of 221 of 291 subjects were included; 4 (1.8%) eyes were classified as definitively having glaucoma. Depending on the reference standard diagnosis as well as on the borderline test-positive definition of the HRTII-3, sensitivity and specificity varied between 36.4% and 100%. For HRTII/MRA, 88.2% to 96.5%, 54.5% to 100%, and 74.5% to 93.6%; for HRT3/MRA, 61.9% to 100% and 64.3% to 85.2%; for HRT3/GPS, 85.7% to 100%; and for combination HRT3/MRA/GPS, 73.4% to 78.2%.

Conclusions: In this pilot study, the MRA of the HRT3 appears to be more sensitive but less specific than the MRA in the HRTII version. The GPS, although somewhat less specific than MRA (and some discs are nonclassifiable by this technique), had greater sensitivity and as a screening method may have the additional advantage of being contour-line independent.