To evaluate both the concordance in the classification of diabetes by clinical and C-peptide criteria and, prospectively, the consistency of the classification by C-peptide.

Individuals with diabetes who were enlisted in the prospective epidemiological study of diabetic neuropathy (Rochester Diabetic Neuropathy Study [RDNS]) were classified clinically by National Diabetes Data Group (NDDG) criteria to IDDM and NIDDM at entry to the study. In addition, C-peptide response to 1 mg glucagon was measured at entry for the classification to IDDM (basal C-peptide, < 0.17 pmol/ml; increment above basal, < 0.07 pmol/ml) and NIDDM (all other responses) and for concordance with the clinical classification made. The consistency of the C-peptide response was assessed every 2 years for up to 8 years.

Among 346 individuals with diabetes, 84 were classified as IDDM and 262 as NIDDM by clinical algorithm. COncordance with the C-peptide response occurred in 89% of the patients and remained consistent during 8 years of follow-up. Among the 37 patients with discordant clinical and C-peptide classification, those considered clinically to have NIDDM had a consistent IDDM C-peptide response during follow-up, and most of those considered to have IDDM clinically eventually showed an IDDM C-peptide response during follow-up.

Clinical criteria for the classification of diabetes are highly correlated with the assessment of insulin secretory reserve. A small number of individuals considered to have NIDDM clinically or by C-peptide have or develop an IDDM peptide response.