Context: Mexican-Americans have more diabetes than non-Hispanic whites, but the extent to which insulin resistance and insulin secretion explain the ethnic difference is unknown.
Objective: We analyzed selected indices of insulin resistance and secretion for both the ethnic difference and predictive discrimination.
Design and setting: The San Antonio Heart Study is a longitudinal population-based study with a follow-up period of 7.5 yr.
Participants: A total of 1540 nondiabetic individuals aged 25-64 yr were enrolled from January 1984 to December 1988.
Interventions: Homeostasis model assessment (HOMA) of insulin resistance and secretion were estimated by available formulas (HOMA-IR and HOMA β-cell) and computer program (HOMA2S and HOMA2B). Matsuda index and insulinogenic index from 0 to 30 and 0 to 120 min (ΔI0-30/ΔG0-30 and ΔI0-120/ΔG0-120) were also calculated.
Main outcome measure: Incident diabetes was defined by the 2003 American Diabetes Association criteria.
Results: Incident diabetes was in excess in Mexican-Americans [odds ratio 2.26 (95% confidence interval, 1.53-3.34)]. Matsuda index explained a larger proportion of the ethnic difference than did HOMA-IR (49.2 vs. 31.0%; P<0.001). The ethnic difference was not explained by measures of insulin secretion. Matsuda index and ΔI0-30/ΔG0-30 had a better predictive discrimination than their HOMA equivalents and ΔI0-120/ΔG0-120. HOMA estimates by the computer program offered no advantage over simple formulas for HOMA.
Conclusions: Insulin resistance accounts for a large and significant proportion of the excess risk of diabetes in Mexican-Americans. Matsuda index is better than HOMA-IR for both explaining the ethnic difference and predicting diabetes.