Purpose: Agreement between three observers--two recently trained fellows and their supervisor--was measured using estimations of cup/disc ratio from stereoscopic optic nerve head photographs and planimetric measurements of cup/disc ratio. Agreement between the clinicians' planimetric measurements of cup/disc ratio and laser scanning tomographic measurements of cup/disc ratio also was assessed.
Methods: From 16 stereoscopic optic nerve head photographs of 16 subjects, the three observers performed clinical estimations of horizontal and vertical cup/disc ratios and planimetric measurements of cup/disc ratios. Interobserver agreement was measured using intraclass correlation coefficients (ICCs). Agreement between the planimetric cup/disc ratios and laser scanning tomographic cup/disc ratios obtained with the Heidelberg Retina Tomograph (HRT; Heidelberg Engineering, Heidelberg, Germany) also was measured using ICCs. The difference between the planimetric and HRT cup/disc ratios was calculated.
Results: The agreement between observers for clinical estimations from stereoscopic optic nerve head photographs (ICC = 0.74 horizontally and 0.83 vertically) was substantial. Agreement between the observers' planimetric measurements of cup/disc ratio was substantial (ICC = 0.79). Agreement between HRT cup/disc ratio and each observer's planimetric cup/disc ratio was moderate (ICC = 0.57-0.65), with large confidence intervals. The cup/disc ratio measured with HRT was an average of 0.07 to 0.11 larger than the planimetric cup/disc ratio.
Conclusion: Substantial agreement between observers can be achieved when estimating cup/disc ratio with stereoscopic optic nerve head photographs and with planimetric measurements of cup/disc ratios, provided there is a standard protocol and sufficient training period. Good agreement is critical in a teaching institution to ensure accurate follow-up care of patients with glaucoma, especially if patients are examined by different clinicians. Laser scanning tomography is a more repeatable and objective method, which may provide further standardization of optic nerve head assessments. Future studies will determine the reference plane that optimizes agreement between the HRT findings and each clinician's estimations.