Supplementary Materials and Methods

Supplementary Materials and methods
Patient Selection
Patients were studied prospectively at Oregon Health & Science
University (OHSU), a tertiary-care university hospital with 72 ICU beds in 4 ICUs,
from May 2004 through November 2006. The protocol was approved by the
hospital institutional review board (IRB00001403) and consent was waived.
Prospective data collections were made from patients who were at least 18 years
of age and receiving IIP for hyperglycemia in the medical, surgical, neurosurgical
and trauma ICUs. Cardiothoracic surgery patients were excluded as they were
treated with a different IIP.
Insulin Infusion Protocol
The IIP has been in use at OHSU since 2002. It is based on the Leuven protocol
and modified by the OHSU ICU oversight committee. The protocol was
recommended for all non cardiothoracic ICU patients if blood glucose was
greater than 120 mg/dL. The protocol defines the starting dose of insulin and the
rates of increase of insulin to control blood glucose. Hourly blood glucose
measurements are required until blood glucose is controlled and the target
glucose is 80-110 mg/dL. Physicians caring for the patients determined whether
two or four hourly glucose measurements were optimal once the glucose values
were stable, and whether corticosteroids were to be given as infusions or bolus.
The bedside nurse controlled the insulin infusion rate based on the protocol.
Human insulin (Novo Nordisk A/S, Denmark) infusions were prepared by the
hospital pharmacy at 1 unit/mL in normal saline.
Data Collection
Blood for glucose measurement was obtained using finger stick or
indwelling catheters by the nurse caring for the patient. Glucose measurements
made using ICU based point-of-care glucose meters that were calibrated daily
(LifeScan, Inc., Johnson & Johnson, Milpitas, CA). Patient charts in four
intensive care units were prospectively reviewed. If the IIP was stopped for >24
hours and then recommenced the second data collection was considered a new
patient episode. Prior to the first data collection period, we audited ICU patient
charts for two week periods during May 2004, August 2004, November 2004, and
June 2005, and determined blood glucose values for patients receiving insulin by
the IIP. Data were collected by two or three team members. Between these
audits, measures were undertaken to improve the proportion of measurements in
target range. These interventions included supplementary nursing and physician
training, performance feedback to nurses and physicians and the purchase of
additional glucose meters.
In 2006, data were collected between 05/05-08/06 and 08/30-10/31 daily
by the volunteers of the Critical Care Academic Associates Program (CCAAP).
The program consisted of ~30 volunteers who work as part of the Center for
Intensive Care Research (CICR) at OHSU to conduct patient screening,
enrollment and data collection. CCAAP volunteers received education on glucose
control in the ICU, the IIP, and the data format for this study. All data collection
was supervised by a CCAAP coordinator and audited for accuracy.
Data Analysis
Glucose measurements were entered into a database and the distributions
compared for each period of data collection. The data were tested for Gaussian
distribution using the Kolmogorov-Smirnov test, GraphPad Prism version 4.03 for
Windows (GraphPad Software, San Diego, CA, USA). In the absence of a
Gaussian distribution we compared different continuous data groups using the
non-parametric Mann- Whitney test. Categorical comparisons were made using
the Chi-square test. Pearson test was used to calculate correlation. All data are
shown as mean ± standard error of mean or mean and range. Curve fitting to
Gaussian or sine functions was performed using a non-weighted least squares
technique using IgorPro (Wavemetrics, Lake Oswego, OR).