Relationship of baseline glucose homeostasis to hyperglycemia during medical critical illness

Abstract

Study objective: To elucidate the relationship of baseline glucose control and acute stimuli with hyperglycemia during medical critical illness.

Design: Prospective cohort study.

Setting: Medical ICU (MICU) of a university affiliated hospital.

Patients: Convenience sample of 100 medical patients meeting criteria for severity of illness and anticipated length of stay and not admitted to the hospital for diabetic ketoacidosis or a hyperglycemic hyperosmolar state.

Interventions: None.

Measurements and main results: Patients were categorized as having normal, abnormal, or unevaluable baseline glucose control based on history and glycosylated hemoglobin (HbA1c). Data collection included blood glucose measurements within 120 h of MICU admission, and dosing of norepinephrine, corticosteroids, propofol, and carbohydrates. Average blood glucose and times over glycemic thresholds were calculated using linear interpolation. Hyperglycemia (glucose > 110 mg/dL) was pervasive in all groups. Among the 51 patients with normal baseline glucose control, HbA1c was correlated with hyperglycemic time (p < 0.01, R(2) = 0.15). Multiple regression found HbA1c, age, corticosteroid dose, and carbohydrate administration independently associated with hyperglycemic time (p < 0.05 for each, total R(2) = 0.49) in these patients, while body mass index, APACHE (acute physiology and chronic health evaluation) II, norepinephrine dose, propofol dose, gender, and sepsis were not associated with time > 110 mg/dL. Among normal subjects, HbA1c was independently predictive of peak and average glucose, and the fraction of time glucose was > 150 mg/dL and > 200 mg/dL (p < 0.05 for each). Patients with abnormal baseline glucose control had significantly more hyperglycemia than patients with normal baseline control.

Conclusions: Even in patients without evidence of abnormal glucose homeostasis at baseline, hyperglycemia is common during critical illness. Time exposure to hyperglycemia is correlated with acute stressors and baseline glucose regulation, as characterized by HbA1c. Patients with low HbA1c levels are less disposed to hyperglycemia during severe illness than patients with higher, but still normal, HbA1c.