Diagnostic accuracy of the Heidelberg Retina Tomograph for glaucoma a population-based assessment

Abstract

Objective: To evaluate the Heidelberg Retina Tomograph II (HRTII; Heidelberg Engineering, Heidelberg, Germany) as a screening tool for open-angle glaucoma (OAG) in an older population.

Design: Population-based, cross-sectional study.

Participants: Participants from the 10-year follow-up Blue Mountains Eye Study (n = 1952; 75.6% of survivors).

Methods: Participants underwent optic nerve head imaging performed using the HRTII; OAG was diagnosed independently from optic disc photographs and Humphrey 24-2 visual fields.

Main outcome measures: The Moorfields Regression Analysis (MRA) was applied to scans using the default results of "normal," borderline," and "outside normal limits." Outcome classification was at the person level and used data from both eyes.

Results: Mean age was 73.7 years. The HRT scans could be acquired in 1644 participants, 95.9% of those fully examined; 87.4% of scans having a topography standard deviation (TSD) < or =40 microm. Larger TSD was associated with older age and OAG. The MRA sensitivity was 64.1%, specificity 85.7%, positive predictive value 21%, and negative predictive value 97.6% for detecting OAG. Including borderline results improved sensitivity (87.0%) but specificity dropped to 70.6%. Significant predictors of abnormal MRA included OAG (odds ratio [OR], 7.70; 95% confidence interval [CI] 4.79-12.35), age (OR, 1.05; 95% CI, 1.03-1.07), TSD (OR, 1.01; 95% CI, 1.003-1.020) and disc size (OR, 3.2; 95% CI, 2.31-4.49). Diagnostic accuracy was improved by restricting TSD <40 microm. Further TSD restriction improved specificity at the expense of sensitivity. As disc size increased, specificity fell whereas sensitivity, OAG prevalence, and the proportion testing positive rose. Glaucoma prevalence, positive predictive value, and the proportion testing positive increased, but specificity fell with increasing age. Sensitivity also fell between youngest and oldest groups. Diagnostic performance improved for visual field mean deviation < -4.0 dB. Single eye analyses overestimated specificity and underestimated sensitivity compared with using data from both eyes.

Conclusions: Diagnostic test evaluation studies of HRTII using single eye analyses or restrictive selection overestimate test accuracy compared with this population-based study. Although the specificity of the MRA was inadequate for use as a glaucoma screening test, the HRTII performed relatively well in an unselected older population with acceptable quality scans in most eyes.