Objective: To investigate duration of the period between diabetes onset and its clinical diagnosis.

Research design and methods: Two population-based groups of white patients with non-insulin-dependent diabetes (NIDDM) in the United States and Australia were studied. Prevalence of retinopathy and duration of diabetes subsequent to clinical diagnosis were determined for all subjects. Weighted linear regression was used to examine the relationship between diabetes duration and prevalence of retinopathy.

Results: Prevalence of retinopathy at clinical diagnosis of diabetes was estimated to be 20.8% in the U.S. and 9.9% in Australia and increased linearly with longer duration of diabetes. By extrapolating this linear relationship to the time when retinopathy prevalence was estimated to be zero, onset of detectable retinopathy was calculated to have occurred approximately 4-7 yr before diagnosis of NIDDM. Because other data indicate that diabetes may be present for 5 yr before retinopathy becomes evident, onset of NIDDM may occur 9-12 yr before its clinical diagnosis.

Conclusions: These findings suggest that undiagnosed NIDDM is not a benign condition. Clinically significant morbidity is present at diagnosis and for years before diagnosis. During this preclinical period, treatment is not being offered for diabetes or its specific complications, despite the fact that reduction in hyperglycemia, hypertension, and cardiovascular risk factors is believed to benefit patients. Imprecise dating of diabetes onset also obscures investigations of the etiology of NIDDM and studies of the nature and importance of risk factors for diabetes complications.