Use of serum and plasma glucose measurements as a benchmark for improved hospital-wide glycemic control

To demonstrate the benefit of an institutionally implemented glucose control intervention based on serum and plasma glucose values in the acute inpatient setting.

In a retrospective analysis, all serum and plasma glucose values from the laboratory information system database from 1999 through 2005 were used to assess implementation of 2 new hospital-wide intravenous and subcutaneous protocols aimed at lowering blood glucose values without increasing the number of hypoglycemic events. In our analysis, we used both a per-patient hyperglycemic index (HGI), an area-under-the-curve analysis, and hospital-wide geometric mean blood glucose to assess glucose control. Bedside capillary blood glucose measurements were not included.

More than 630,000 serum and plasma glucose results were available for analysis. The percentage of results above the protocol target of 180 mg/dL decreased from 16.4% before the intervention to 10.0% after the intervention (P<.00001), and we found no change in the proportion of "critical" hypoglycemic results (<50 mg/dL). The hospital-wide geometric mean decreased significantly and coincided with a significant decrease in the fraction of patients with poor glucose control (based on the HGI) from 27.6% to 18.7% (P<.00001). The geometric mean blood glucose was found to be an excellent marker for the HGI (r2 = 0.99).

We are the first to report improvements in glucose control over an extended period with use of both hospital-wide intravenous and subcutaneous insulin protocols in an academic hospital setting. Furthermore, hospital-wide mean blood glucose levels are excellent surrogates for the more comprehensive calculation of per-patient HGI.