Abstract
Purpose: :
The diagnosis of CSME is based on three criteria: central thickening within 500 µm from the fovea, hard exudates within 500 µm from the fovea with associated retinal thickening, retinal thickening at least one disc area in size, part of this within one disc diameter from the fovea. The definition has been used for decades and treatment strategies are based upon it. With increasing use of objective, quantitative measurements of retinal thickness a correlation of the clinical definition and OCT is needed. Thus we sought to evaluate if retinal thickness calculated with optical coherence tomography (OCT) is able to discriminate clinically significant macular edemas (CSME) from no or non–significant edemas (NCSME).
Methods: :
108 eyes from healthy controls, 56 eyes from diabetic patients with CSME and 66 eyes from diabetic patients with no or NSCME were examined with OCT3/Stratus and the thickness of the nine fields defined by ETDRS were recorded. Healthy and CSME eyes were evenly randomised into a calibration set and a test set, where classification on the test set is made based on the weights from the calibration set. Linear discriminant analysis was used to identify CSME, taking into consideration a difference in the number of patients and controls. In a second analysis the healthy eyes of the test set was substituted by diabetic patients with no or non–significant edema.
Results: :
The calibration set classified all 54 healthy eyes correctly as well as 26 out of 27 eyes with CSME, thus the specificity was 100% and the sensitivity 96%. In the test set all 54 healthy eyes were correctly classified, while 23 of 29 eyes with CSME were correctly classified yielding a specificity of 100% and a sensitivity of 79%. If the healthy eyes of the test set were substituted with eyes with NCSME, similar results were found.
Conclusions: :
The diagnosis of CSME from clinical examination is subjective and easily biased. With objective OCT, a direct interpretation of retinal thickness comparable to the clinical diagnosis is feasible.
Keywords: diabetes • macula/fovea • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology