[Inter-observer variability in the diagnosis and classification of diabetic retinopathy through biomicroscopy]

Analysis of the inter-observer variability of biomicroscopy used for the diagnosis of Diabetic Retinopathy.

This was a descriptive study. Parallel observer-blind evaluations of the degree of retinopathy in type 2 diabetic patients, as defined on biomicroscopic photographs, were performed by two ophthalmologists. The sample size required for the Kappa index among ophthalmologists with a disagreement ratio of 15%, precision ratio of 5% and confidence level of 95% is n=196 (<> being the number of eyes). The only variable measured was the degree of Diabetic Retinopathy, according to the modified Early Treatment Diabetic Research Study (ETDRS) classification.

The average age of the 217 patients photographed was of 65.42 years (SE= 9.91). In 191 instances there was total agreement between the 2 ophthalmologists. In 24 instances the discrepancy was only of one degree of the classification of the ETDRS, and in 2 the discrepancy was of two degrees. In no case was it greater than this. (Quadratic weighed Kappa = 0.876, IC95%: 0.655-0.952 and linear weighed Kappa = 0.804, IC95%: 0.729-0.878).

The Weighed Kappa index demonstrated a <> agreement of the degree of diabetic retinopathy. The discrepancies were slight, were of no clinical importance, and would not have affected treatment decisions. The results indicate that this examination, performed by a single ophthalmologist, can be utilised as a reference standard in Diabetic Retinopathy diagnosis.