A study from the POAAGG highlights the limitations of standard OCT normative databases in classifying RNFL thickness for individuals of African ancestry. The analysis shows that an ancestry-specific model results in fewer abnormal classifications compared to the Cirrus HD-OCT database.
Background
Optical coherence tomography (OCT) is crucial in glaucoma management, providing color-coded reports that indicate retinal nerve fiber layer (RNFL) thickness. The accuracy of these classifications is dependent on the normative databases used, which may not adequately represent diverse populations. This study emphasizes the need for more representative datasets to avoid misclassification of healthy eyes as abnormal.
Data Highlights
Classification
Cirrus Database
POAAGG Reclassification
Red
27.7%
24.4%
Green
51.0%
56.7%
Key Findings
The ancestry-specific model produced fewer abnormal classifications than the Cirrus database.
12.8% of RNFL measurements initially classified as red by Cirrus were reassigned to yellow or green.
23.0% of yellow classifications from Cirrus shifted to green under the POAAGG model.
In the temporal quadrant, 80.3% of red codes moved to yellow.
Among stage 1 eyes, 19% of average red codes shifted towards greener classifications.
Clinical Implications
Clinicians should recognize that OCT color codes are probability labels rather than definitive diagnoses.
Conclusion
The findings emphasize the importance of using appropriate normative databases in OCT analysis.
Justin Schweitzer, OD, FAAO, and Jessica Steen, OD, FAAO, describe emerging technologies, treatment strategies, and clinical considerations that can help eyecare professionals (ECPs) identify glaucoma earlier and better manage progression.
