advertisement
It is common clinical knowledge, that detecting advanced glaucoma is a relatively easy clinical task. On the other hand, detection of the disease at an early stage can be challenging. Leite et al. (1205) evaluated if this holds true to spectral domain optical coherence tomography (SD-OCT) diagnostic ability. One hundred thirty five healthy and glaucomatous eyes with a wide range of disease severity were enrolled. Disease severity was defined by visual field (VF) global indices. Mean baseline VF mean deviation in the glaucomatous eyes was -5.63dB. The best retinal nerve fiber layer (RNFL) parameter to discriminate between healthy and glaucomatous eyes, as determined by the area under the receiver operating characteristics (ROC), was the mean RNFL (0.892) followed by the RNFL thickness in the inferior and superior quadrants. Using ROC regression models the authors showed that disease severity was a statistically significant variable in the model. This implies that for a more severe disease, a higher area under ROC can be expected compared with earlier stages of glaucoma. It is interesting to note that for very early disease, inferior quadrant RNFL thickness outperformed the mean RNFL though it was not reported if this difference was statistically significant. The study confirms the clinical observation and provides a tool to compare ROCs across studies with different disease severity popula- tions. However, due to the small sample of subjects with advanced disease the accuracy of the estimate at this range might be limited.