Aim: HOMA and QUICKI are the most widely used indices for assessing insulin sensitivity. Both are based on fasting glucose and insulin measures, and mainly differ by the log transformation of these variables in QUICKI. However, HOMA is less reproducible than QUICKI, and log HOMA does not improve its reproducibility. The aim of this study was to investigate the various mathematical transformations of HOMA and to assess its reproducibility.
Method: We used data from a clamp study involving 123 non-diabetic overweight and obese postmenopausal women. Fasting insulin and glucose were measured in two visits 15 and 30 days apart. This allowed us to calculate HOMA as (fasting glucose [mmol/L] x fasting insulin [microU/mL])/22.5 and QUICKI as 1/(log fasting glucose [mg/dL]+log fasting insulin [microU/mL]) twice for subjects who were weight-stable between visits.
Results: QUICKI had better reproducibility (CV=3.9%) than either HOMA (CV=26.7%) or log HOMA (CV=22.0%). However, log-transforming HOMA using log (glucose x insulin)/log (22.5) and log-transforming HOMA without transforming the constant denominator improved its CV to 6.5% and 5.7%, respectively.
Conclusion: By modifying the mathematical expression of HOMA, we were able to achieve comparable CVs for QUICKI and HOMA. However, the CV should be used to assess the reproducibility of techniques to measure glucose and insulin, not of mathematical formulas. When evaluating indices for the assessment of insulin sensitivity, the key point is how well they correlate with the 'gold-standard' glucose clamp.