Prediction Models for Insulin Resistance

A prediction model for estimating insulin resistance in hypertensive patients is presented. Body-mass index, serum triglyceride concentrations and liver enzyme activity in plasma correlate to insulin resistance determined with the euglycaemic, hyperinsulinaemic clamp technique. Prediction models using body-mass index and either triglycerides or serum alanine-amino transferase were equally good in predicting insulin resistance and gave results that were as reliable as those obtained in a model using fasting-insulin concentrations. The hyperinsulinaemic clamp had a reproducibility error of 14%, and body-mass index and serum triglycerides had a multiple correlation of 0.57 to the insulin-sensitivity results. The model predicts insulin resistance with acceptable statistical power, whereas the power to predict high values of insulin sensitivity is less good.