Limitations and problems of diabetes classification from an epidemiological point of view.

There are residual ambiguities between the two main current glycaemic definitions of the categories of DM, IGT and normal GT which should be resolved. IGT is clearly a highly heterogeneous category and could with advantage be resolved into its identifiable subsets though adequate data for this is not yet available. The concept of insulin dependency requires clearer definition for operational purposes. Biochemical parameters (e.g. C-peptide responses) may help. Attempts to combine clinical manifestations and pathogenic mechanisms in a single classification (e.g. IDDM/NIDD versus Type I/Type II) should be handled with care. If the term Type I is to be retained, it should be applied to a defined pathogenic process, not to a clinical type of DM. The term Type II is inadequately defined at present. IDDM and NIDDM, clinical descriptive terms, may be provoked by a variety of pathogenic mechanisms (i.e. they are 'heterogeneous'). They could be subclassified by mechanism (when known). More visibility should be given in classification to non-Europid forms of DM (e.g. 'Tropical or 'Nutritional' DM). A staging dimension should be recognised in classifications of DM. Future classifications will benefit from the incorporation of the presence or absence of susceptibility/resistance factors to diabetes itself or to its severe long term sequelae. There remain uncertainties about the definitions and clinical implications of gestational DM (and gestational IGT) not discussed above. It should be accepted that different user groups may need different subclassification of diabetes and glucose intolerance to meet their specific requirements and so long as this is made clear and definitions are adequate this should not be a problem. However, for the present, all groups should accept the proposed glycaemic definitions of DM or IGT for the purposes of comparability.