Consumer continuous glucose monitoring uncovers hidden metabolic risk in free-living adults without diabetes

Whether continuous glucose monitoring (CGM) provides actionable information in adults without diabetes is contested, as its metrics were designed for diabetes and screening relies on HbA1c. In this cross-sectional study of 62,122 free-living adults on the Levels digital-health platform, average glucose was the most reliable measure across weeks (intraclass correlation 0.747). Among 7,930 participants without diabetes with laboratory data, it tracked insulin resistance, triglycerides, HDL cholesterol, uric acid, and C-reactive protein more closely than HbA1c, explaining nearly three times more variance in fasting insulin (R² 0.131 vs 0.048). With a normal HbA1c, occult hyperinsulinemia (fasting insulin ≥10 µIU/mL) was twice as common at average glucose ≥110 mg/dL (27.5% vs 12.6%). Counterintuitively, at the same average glucose, spike width, not height or frequency, tracked insulin resistance, triglycerides, and C-reactive protein; carbohydrate sensitivity, although reliable, did not track insulin resistance. This analysis suggests a framework for interpreting CGM in adults without diabetes.