PURPOSE To compare evaluation by clinical examination with image grading at a reading center for the classification of diabetic retinopathy and diabetic macular edema.
METHODS Action to Control Cardiovascular Risk in Diabetes (ACCORD) and Family Investigations of Nephropathy in Diabetes (FIND) had similar methods of clinical and fundus photograph evaluation. For analysis purposes, the photographic grading scales were condensed to correspond to the clinical scales, and agreement between clinicians and reading center classification were compared.
RESULTS Six thousand nine hundred and two eyes of ACCORD participants and 3,638 eyes of FIND participants were analyzed for agreement (percent, kappa) on diabetic retinopathy on a 5-level scale. Exact agreement between clinicians and reading center on diabetic retinopathy severity category was 69% in ACCORD and 74% in FIND (kappa 0.42 and 0.65). Sensitivities of the clinical grading to identify the presence of mild nonproliferative retinopathy or worse were 0.53 in ACCORD and 0.84 in FIND. Specificities were 0.97 and 0.96, respectively. Diabetic macular edema agreement in 6,649 eyes of ACCORD participants and 3,366 eyes of FIND participants was similar (kappa 0.35 and 0.41). Sensitivities of the clinical grading to identify diabetic macular edema were 0.44 and 0.53 and specificities were 0.99 and 0.94, respectively.
CONCLUSION The results support the use of clinical information for defining broad severity categories but not for documenting small-to-moderate changes in diabetic retinopathy over time.