Purchase this article with an account.
P. R. Healey, P. Mitchell, Blue Mountains Eye Study; The Glaucoma Likelihood Score: A Population-Based Model for Glaucoma Screening. Invest. Ophthalmol. Vis. Sci. 2008;49(13):5054. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To assess the utility of a simple screening algorithm for open-angle glaucoma using ocular history and examination in comparison to current diagnostic algorithms.
Screening models were developed using multiple logistic regression to calculate diagnostic odds ratios (OR) and receiver operating characteristic (ROC) curves. Covariate selection utilised previously identified risk factors and signs ascertained from participants from the Blue Mountains Eye Study. Analysis of rank scoring was used to create simple decision rules, which were then compared to the regression models. Age-specific rates of positive screening from the final diagnostic model were used to calculate nationwide impact.
The best estimate of the current diagnostic pathway had a sensitivity and positive predictive value of 49%. Almost all glaucoma cases with lower intraocular pressure (IOP) or younger age were undiagnosed prior to the study. IOP performed poorly as a test for open-angle glaucoma with an area under the ROC curve of 0.722. Vertical cup-disc ratio performed quite well with an ROC area of 0.96. In multivariate diagnostic modelling, the best cut-off values for intraocular pressure and cup-disc ratio were 21mmHg and 0.7 respectively. The simplified empirical model (Glaucoma Likelihood Score) was based on addition of the following factors; twice the age in decades, 10 points for any eye IOP>21mmHg, 12 points for a disc haemorrhage, 3 points for myopia <= -1D, 30 points for any vertical cup-disc ratio >0.7 and 45 points for any notch of the disc rim. The final model had an ROC area of 0.982. A cut-off of 50 points (GL50) would screen 4.5% of the population over 50 years of age positive, with a sensitivity of about 90% and a positive predictive value of 56%. Projecting these to the Australian population, the GL50 screening algorithm would detect almost all of the currently undiagnosed open-angle glaucoma cases, particularly in the younger age groups.
The GL50 screening test for open-angle glaucoma is simple, inexpensive and accurate. It has the best diagnostic screening test results ever reported for glaucoma. If verified in another population, it has the potential to become a useful tool in glaucoma diagnosis.
This PDF is available to Subscribers Only