Evaluation of the performances and costs of a spectrum of dynamic insulin sensitivity test protocols

The strategic design of most insulin sensitivity (SI) tests maximises either accuracy or economy, but not both. Hence, accurate, large-scale screening isn´t feasible. The dynamic insulin sensitivity test (DIST) was developed to better optimize both important metrics. The highly flexible DIST protocol samples insulin, glucose and C-peptide during a comparatively short test. Varying the sampling periods and assays, and utilising alternative computational methods enables a wide range of tests with varied cost and accuracy tradeoffs. The result is a hierarchy of tests to facilitate low-cost screening. Eight variations of the DIST are evaluated against the fully-sampled test by correlating SI and endogenous insulin production (Uen(t)) metrics. Five variations include sample and assay reductions and three utilise quick-DIST (DISTq) parameter estimations. The DISTq identification methods only require glucose assays and thus enable real-time analysis. Three DISTq methods were tested; the fully-sampled, the Short, and the 30 minute two-sample protocol. 218 DIST tests were completed on 84 participants to provide the study data. Methods that assayed insulin replicated the findings of the full DIST particularly well (R=0.89~0.92) while those that assayed C-peptide managed to best replicate endogenous insulin metrics (R=0.72~1.0). The three DISTq protocols correlated to the fully-sampled DIST at R=0.83, 0.77 and 0.71 respectively. As expected, test resolution increased with rising protocol cost and intensity. The ability of significantly less expensive tests to replicate the values of the fully-sampled DIST was relatively high (R=0.92 with four glucose and two insulin assays and 0.71 with only two glucose assays). Thus, an SI screening programme could achieve high resolution at a low cost by using a lower resolution DIST test. Whe