Sliding-scale Insulin Does Not Control Steroid-induced Hyperglycemia in a Non-diabetic Patient, No Matter How You Tune It

Abstract

Sliding-scale correction insulin is the standard hospital response to high blood glucose. Its constants were derived in people who secrete no insulin, yet it is routinely ordered for patients whose own secretion is intact. We represent the correction rule as a proportional controller with a deadband, sampled four-hourly, acting on a glucose model built from published physiology, including pancreatic feedback. We derive the sensitivity of glucose to exogenous insulin, evaluate it in 4,995 virtual patients spanning published parameter ranges, and calibrate the model to a non-diabetic adolescent receiving high-dose corticosteroids and parenteral nutrition, with 45 bedside glucose measurements. Here we show that endogenous regulation restores glucose with a time constant of 11.5 minutes, against 64 minutes for injected insulin to peak, so most of each dose replaces the patient's own secretion and the ordered rule moves mean glucose by less than 1.5 mg/dL. The ordered correction factor exceeds per-unit potency in every virtual patient. Because potency falls as glucose rises, in 99.5% of virtual patients no constant correction factor is both effective at 190 mg/dL and free of overshoot at 110 mg/dL; the mismatch exceeds 25% in 79% and grows with the glucose of half-maximal secretion. Four-hour sampling is phase-locked to dextrose in medication diluents and misses the resulting excursions. When endogenous secretion is intact, the sliding scale fails for structural reasons that retuning cannot fix. A continuous insulin infusion co-delivered with the dextrose and adjusted once daily on mean glucose matches what four-hour monitoring can measure.