Abstract
Purpose: :
To evaluate the Heidelberg Retinal Tomograph (HRT) for detecting open-angle glaucoma (OAG) in an older Australian population.
Methods: :
The HRT II imaged participants in the Blue Mountains Eye Study 10-year follow-up (BMES III). The Moorfields regression analysis (MRA) assessed optic disc scans as "normal", "borderline" and "outside normal limits". OAG was diagnosed independently using optic disc photographs and 24-2 Humphrey visual field printouts.
Results: :
The mean age of the BMES III cohort was 73.7 years. HRT scans were acquired in 1607 participants, 91.3% of those examined. Eighty-eight percent of scans had a topography standard deviation (TSD) of 40 micrometers or less, with 52% having a TSD of 20 micrometers or less. Increasing TSD was associated with older age and the presence of open-angle glaucoma (OAG). OAG was detected in 105 participants - the overall glaucoma prevalence of 5.43% was identical to that in BMES I, after accounting for the older age of participants in BMES III. Of the 1607 participants who had HRT scans, the MRA result was normal in both eyes of 1080 participants (67.2%). At least one eye was borderline in 252 participants (15.7%) and abnormal in 275 participants (17.1%). This third group included 184 left and 166 right eyes, and in 75 participants, both eyes were classified as abnormal. The MRA had a sensitivity of 64.0% (confidence interval, CI, 47.6-52.4), a specificity of 85.7% (CI 84.0-87.4), a positive predictive value of 21.5% (CI 19.5-23.5) and a negative predictive value of 97.5% (CI 96.7-98.3). The positive predictive value increased with age, from 11% for ages 60-69 years to 20% for 70-79 years and 27% for 80 years or older. If borderline results were also classified as abnormal, the sensitivity improved to 87% (CI 85.4-88.6%), but specificity dropped to 71% (CI 68.8-73.2). The positive predictive value fell to 15% (CI 13.3-16.8) while the negative predictive value improved to 98.9% (CI 98.4-99.4). Significant predictors of an abnormal MRA were OAG (odds ratio, OR 8.3, CI 5.2-13.3), older age (OR 1.05 per year CI 1.03 - 1.07), TSD (1.011 per micrometer CI 1.004 - 1.018) and larger optic disc size (OR 5.4 per micrometer of vertical disc diameter, CI 2.3 - 12.5). Restricting analysis to scans with better TSDs had little impact on the diagnostic accuracy of the MRA. Analysis by disc size suggested that the MRA would over-diagnose abnormality in large discs.
Conclusions: :
The HRT II acquired acceptable scans in the majority of eyes in this unselected older population, including those with significant lens opacity. As a screening test for open-angle glaucoma, the MRA performed little better than tonometry or visual field screening.
Keywords: clinical (human) or epidemiologic studies: risk factor assessment • optic disc • imaging/image analysis: clinical