Emergency Department (ED) patients with severe hyperglycemia will be randomized to two treatment goals: discharge glucose less than 600 mg/dL or less than 350 mg/dL.

Randomization is stratified by whether the chief complaint is "High Blood Sugar" in the electronic medical record or other.

It is common practice to provide insulin and/or intravenous (IV) fluids to lower glucose levels prior to discharge in patients that present with hyperglycemia. There is, however, no data supporting this practice.

A recent retrospective observational cohort study (HSR# 12-3497) of 567 ED patients with severe hyperglycemia demonstrated no association between discharge glucose levels and short-term adverse outcomes at 7 days (return ED visits, diabetic ketoacidosis (DKA), hospitalization, and death).

There were no other short-term adverse outcomes associated with the degree of glycemic control. Less than 1% of patients developed DKA at 7 days. No patients died.

As this was a retrospective study, the conclusions that can be drawn are limited by the lack of randomization and control; it is necessary to conduct a randomized controlled trial to quantify how much time is expended caring for these patients, and gather safety data.

This is important because much time and money is focused on glucose reduction, which may not change short-term outcomes. Furthermore, in retrospective analysis approximately 1.5% of patients receiving therapy for hyperglycemia developed iatrogenic hypoglycemia during the ED stay.

To detect a difference in length of stay by 60 minutes (Standard deviation is 100 minutes) with an p=0.05 and b=0.8, enrollment of 45 patients per group will be required, with a total enrollment of 90 patients. Enrollment will stop when 45 patients in each group have been contacted in follow-up.

This study will only include patients who are discharged from the ED. Patients who are enrolled and then eventually admitted to the hospital will be tabulated but excluded from the length of stay analysis.

Statistical analysis will be completed for the primary outcome with a Mann-Whitney U test, assuming the data will not be normally distributed. Secondary outcomes will be compared with chi-squared testing. A secondary regression model will be formed, based on covariates that could influence ED length of stay, such as amount of diagnostic testing completed.

Inclusion Criteria:

• Glucose of more than 400mg/dL but less than 600 mg/dL at any point in the ED
• Working phone number and willing to discuss health status at 7-10 days via phone

Exclusion Criteria:

• Plan for hospital admission
• Already received insulin during the ED stay
• Type 1 diabetes
• Less than 18 years old
• Non-English speaking
• Altered mental status or encephalopathy (unable to provide informed consent)
• DKA, as determined clinically by the treating physician, without a lab requirement.
• Critically ill, as determined by the treating physician.
• Unable to provide informed consent
• Prisoners
• Pregnant women