Characterizing Glucose Changes Antecedent to Hypoglycemic Events in the Intensive Care Unit

ObjectiveTo determine whether patterns of glucose changes before hypoglycemia vary according to the severity of the event.MethodsIn this retrospective analysis, point-ofcare blood glucose (POC-BG) data were obtained from the intensive care units (ICUs) of a convenience sample of hospitals that responded to a survey on inpatient diabetes management quality improvement initiatives. To evaluate POC-BG levels before hypoglycemic events, data from patients who experienced hypoglycemia during their time in the ICU were examined, and their glucose changes were assessed against a comparison group of patients who achieved a glycemic range of 80 to 110 mg/dL without ever experiencing hypoglycemia. Absolute glucose decrease, glucose rate of change, and glucose variability before hypoglycemic events (< 40, 40-49, 50-59, and 60-69 mg/ dL) were calculated.ResultsA total of 128 419 POC-BG measurements from 2942 patients in 89 ICUs were analyzed. Patients who experienced the most severe hypoglycemic episodes had the largest absolute drop in their glucose levels before the event (P < .001). The glucose rate of change before a hypoglycemic event increased with worsening hypoglycemia: mean (± standard deviation) glucose rate of change was-1.69 (± 2.98) mg/dL per min before an episode with glucose values less than 40 mg/dL, -0.56 (± 2.65) mg/dL per min before an episode with glucose values 60 to 69 mg/dL, but only -0.39 (± 0.70) for patients who attained a glucose range of 80 to 110 mg/dL without hypoglycemia (P < .001). Glucose variability before an event progressively increased with worsening biochemical hypoglycemia and was least among patients achieving glucose concentrations in the 80 to 110-mg/dL range without hypoglycemia (P < .001).ConclusionsAntecedent glucose change and variability were greater for patients who experienced hypoglycemia. If monitored, these patterns could potentially alert clinicians and help them take preventive measures. Further examination of how these parameters interact with other predisposing risk factors for hypoglycemia is warranted. (Endocr Pract. 2012;18:317-324)