A BREIF ORIGINAL CONTRIBUTION

Both abnormal insulin levels and low insulin sensitivity have been implicated as risk factors for Type II diabetes mellitus and cardiovascular disease. While insulin level is relatively simple to assess, direct measurement of insulin sensitivity is much more invasive, costly, and time-consuming. The authors considered eight previously described measures or indices of insulin sensitivity derived from the frequently sampled intravenous glucose tolerance test (FSIGT). Each one was evaluated by strength and consistency of association with insulin sensitivity computed from glucose clamp (S1(clamp)), across three glucose tolerance groups, including participants with normal glucose tolerance (n = 11), impaired glucose tolerance (n = 20), and non-insulin-dependent diabetes mellitus (n = 24). Minimal model analysis (MINMOD S1(22)), based on the 22-sample FSIGT, performed best based on statistical criteria of strong and consistent association with S1(clamp). An insulin sensitivity measure similar to that of Galvin et al. (Diabetic Medicine 1990;9:921–8), defined as glucose disappearance (10–50 minutes) divided by insulin area under the curve above baseline from 0–50 minutes, performed best based on statistical criteria and time-savings. Galvin insulin sensitivity is simple to calculate, requires only a 50–minute FSIGT, and is significantly (p ≤ 0.001) and not inconsistently (p = 0.12 for inconsistent association) associated with S1(clamp) over a wide range of glucose tolerance. Am J Epidemiol 1995;142:724–32.