Simplest screen
A 2-hour plasma insulin above 30 µU/mL after a 100 g OGTT identifies the hyperinsulinaemic individual; below 30 µU/mL hyperinsulinaemia is unlikely — proposed as a diagnostic algorithm from the Kraft database (4,185 tests).
Oral glucose · insulin assay · pattern classification
Joseph R. Kraft's 1975 analysis of 3,650 oral glucose tolerance tests showed that insulin curves — not glucose alone — reveal carbohydrate intolerance years before glucose fails. This guide organizes his five patterns into sortable criteria, decay shapes, and physiological interpretation, with a toggle between his original definitions and the adaptations used in modern metabolic health practice.
Kraft's original definitions from Detection of Diabetes Mellitus In Situ (Occult Diabetes), Laboratory Medicine 1975;6(2):10–22 — 100 g OGTT with insulin at fasting, ½, 1, 2 and 3 h (± 4–5 h), in µU/mL.
Click any column header to sort. On small screens, swipe the table sideways. Curves are conceptual shapes illustrating peak timing and decay — not drawn to scale.
The five patterns are distinguished as much by how fast insulin falls as by how high it rises. Kraft graded the return leg with the 2 h + 3 h sum; modern research operationalizations express it as the 3-hour value as a percentage of the 30-minute peak (below 20% normal · 20–65% delayed decline · above 65% delayed peak).
A 2-hour plasma insulin above 30 µU/mL after a 100 g OGTT identifies the hyperinsulinaemic individual; below 30 µU/mL hyperinsulinaemia is unlikely — proposed as a diagnostic algorithm from the Kraft database (4,185 tests).
Both eras flag low-carbohydrate preparation as a false Pattern V: a flat, low insulin curve with normal glucose can simply reflect a low-carb diet. Kraft required a repeat after two or more weeks of high-carbohydrate intake; modern guides ask for roughly 3 days of normal eating with adequate carbohydrate before testing.
Kraft graded glucose with the Wilkerson point system: fasting ≥ 130, 1 h ≥ 195, 2 h ≥ 140, 3 h ≥ 130 mg/dL; two or more points = diabetes. Patterns II–IV with zero points were his “occult” / in situ diabetes.
Insulin thresholds are in µU/mL, identical to mIU/L. Kraft’s series used insulin assays simultaneous with glucose on the same draws; assay methods have changed since, so absolute cut-offs are protocol-dependent.