Predicting diabetes. Moving beyond impaired glucose tolerance

Area of Science:

• Epidemiology
• Biostatistics
• Metabolic Diseases

Background:

• Type II diabetes is a growing public health concern.
• Accurate prediction of type II diabetes risk is crucial for early intervention.
• Existing predictive methods may have limitations.

Purpose of the Study:

• To develop and evaluate multivariate predictive models for type II diabetes.
• To compare the performance of these models against impaired glucose tolerance.
• To identify key risk factors for type II diabetes.

Main Methods:

• Stepwise multiple logistic regression analysis was used.
• A cohort of Mexican Americans and non-Hispanic whites was followed for 8 years.
• Models were developed using demographic, anthropometric, metabolic, and hemodynamic variables.

Main Results:

• Multivariate models demonstrated high predictive accuracy, with relative risks ranging from 12.16 to 35.29.
• Model sensitivity (67.7–83.3%) exceeded that of impaired glucose tolerance (56.5–62.1%).
• Predictive models identified a complex interplay of metabolic and hemodynamic factors.

Conclusions:

• Multivariate predictive models are effective for identifying individuals at high risk of type II diabetes.
• These models offer an alternative to oral glucose tolerance testing.
• The findings underscore the complex syndrome preceding type II diabetes development.