ADVANCING
YOUR PRACTICE
Guardian
of glucose: An electronic algorithm for glycemic control
By Craig Harris, RN, BSN, MPH,
and Dawn Greene, RN, CCRN, CSC
Hyperglycemia is strongly correlated
with increased mortality and increased cost per patient in the
postcardiac surgery population. Nomograms used for optimal glycemic
control haven't proven reliable. Our trial of a software-based,
logarithmic equation program demonstrated optimal glycemic control
and time to control in this patient population.
Background
Hyperglycemia has long been recognized as an inhibitor of the
normal immune response and healing process. Noteworthy studies,
namely the Portland Diabetic Project, have demonstrated that hyperglycemia,
not diagnosis of diabetes mellitus, is the major causal factor
for increased mortality due to complications such as infection
and pump failure.1,2 Moreover, the project demonstrates a direct
correlation between increased cardiac-surgery mortality and a
3-day average capillary blood glucose (CBG) value of more than
150 mg/dL, and is an independent predictor of increased length
of stay in this population; between 0.45 and 0.65 days depending
on the level of hyperglycemia. Other research following prospective
and cohort case studies involving cardiothoracic surgery patients
demonstrated that diabetes and postoperative hyperglycemia were
independently associated with the development of surgical site
infections (SSIs).3
Hyperglycemia is a significant issue
for any ICU patient, not just the postcardiothoracic surgery patient.
In a study of 1,548 patients of a 56-bed predominantly surgical
ICU, researchers found that metabolic control was the strongest
contributor to patient outcomes.4 Their analysis indicated that
the lowered blood glucose (rather than the insulin dose) was a
statistically significant related factor to reduced mortality.
Although we're aware of the importance
of glycemic control in our postsurgical patients, national organizations
that focus on patient outcomes are advocating for tighter glucose
control. The Institute for Health Improvement has identified glycemic
control (between 80 mg/dL and 110 mg/dL in critically ill patients)
as an improvement measure.5 The American Hospital Association
has specified glucose control in the postcardiac surgery patient
by 6 a.m. on postoperative day 1 as a Surgical Care Improvement
Project measure.6
Licensed for 800 beds, North Carolina's
Mission Hospitals, offers award-winning cardiac care. Solucient
named Mission's Heart Program as a Top 100 Heart Hospital for
the years 2000 to 2005 and again in 2007. The Society of Thoracic
Surgeons gave the program its highest rating—three stars.
The cardiovascular recovery unit
(CVRU) at Mission Hospitals is the postanesthesia care unit (PACU)
for the cardiovascular OR. Annually, approximately 1,000 patients
recover in this unit. Postoperative (postoperative day 1 for post-open
heart patients) patients are transferred from the CVRU to the
cardiovascular progressive care unit or to the cardiovascular
ICU (CVICU) if they require ICU care beyond the PACU phase.
We've historically had difficulty
achieving optimal blood glucose control (CBG measurement of 71–150
mg/dL per the Portland Diabetic Project) in our postoperative
cardiac and vascular surgery population. Of the average 1,000
patients treated annually in the CVRU, we were able to achieve
optimal control in only 17%, with an average of 9 hours required
to achieve control using a nomogram-based order set for continuous
insulin infusion titration that's designed specifically for the
ICU patient. We observed serial CBG measurements of greater than
150 mg/dL in 80% of our patients and a 3% rate of hypoglycemia
(i.e., CBG measurement less than 70 mg/dL).
Our trial
We considered several software programs to assist us with glucose
control. Collaboratively with nursing, pharmacy, endocrinology,
CV surgery, and information technology, we chose one particular
software-based tool to trial. Our choice program is a patent-pending,
Health Insurance Portability and Accountability Act (HIPAA)-compliant
program that derives the patient's unique insulin resistance using
a logarithmic equation. The result is continuous insulin infusion
titration recommendations and supplemental insulin (also known
as “sliding scale”) recommendations unique to the
patient.
We implemented a 2-month trial from
March to May 2007. The only criteria for tool utilization were
two or more capillary blood glucose measurements of more than
150 mg/dL. During this time period, 95 of 198 CVRU patients met
these criteria and were managed utilizing the tool. Tool use was
continued upon transfer to the CVICU while on continuous insulin
infusion.
We experienced a dramatic decrease
in CVRU control time from 9 to just over 2 hours. Although we
hadn't previously collected data on control time in the CVICU,
we observed this to be 3.8 hours (which seemed anecdotally a significant
decrease). We also observed our percentage of CBGs within the
optimal range (71–150 mg/dL) to be 80% in the CVRU and 91.5%
in the CVICU—a significant change from our prior 17%. The
incidence of hypoglycemia also decreased significantly, from an
average rate of 3% prior to the trial to less than 1% during the
trial.
Added benefits
In addition to the obvious clinical benefits, simplicity of use,
the ability to customize based on service line/patient population,
and user support are significant benefits. The tool is simple
to use. A basic profile is established for the patient that includes
administration of steroids, the patient's serum creatinine, and
current insulin infusion rate. Based on this information, the
tool recommends an infusion rate and specifies time duration until
next CBG measurement. Each subsequent CBG is entered into the
software, current insulin infusion rate is verified, and additional
calories (such as dextrose infusions) are administered, if applicable.
The software then recommends insulin titration and specifies the
time increment for subsequent CBG measurement. The software alarms,
notifying staff that a CBG is due, and the alarm can't be silenced
until the CBG measurement is entered. This prevents staff from
“overlooking” a CBG and, thus, prevents hyperglycemic/hypoglycemic
spikes.
Training for use of the tool is minimal.
Since we utilize an electronic medical record, staff members were
accustomed to working with electronic patient-care documentation.
The manufacturer provided on-site training. The tool provides
administrative reports that simplify data analysis for performance
improvement and labor studies. The software provides for customization
of insulin management allowing different service lines to tailor
their control and insulin regimens based upon their unique patient
population.
Disadvantages
Since the software customizes control to the individual patient,
frequent CBGs (every hour or more) are often required until control
is achieved. This, along with the insulin titrations associated
with the frequent CBG measurements, significantly increases initial
nursing time.
Compatibility with existing electronic
medical record systems and application delivery systems can be
an issue. There's no interface between the tool and our electronic
medical record and application delivery system. This requires
that nursing staff members manually enter values into the record
and document recommendations using a printed label (which the
software generates). The lack of interface with our application
delivery system restricted software deployment in each patient
room unless manually installed. Therefore, we deployed the software
on key computers throughout the unit.
Posttrial validation
Following the trial, we resumed use of our standard, nomogram-based
insulin order sets. We recollected CBG data to validate our results
and found a return to our pretrial baseline. Optimal blood glucose
control was achieved in only 14% of cases (which was down even
from our prior rate of 17%), along with a 4% rate of hypoglycemia,
an 84% rate of hyperglycemia, and increased control time in our
CVRU patients.
Setting a standard
Our trial data provided overwhelming evidence that the tool improves
glycemic control and decreases control time in this patient population.
In November 2007, we implemented the tool as our standard for
continuous insulin infusion in the CVRU and CVICU.
Our hospital uses an acuity-based
software system for recommended staffing based on projected workload.
Using the indicators within this system, we're able to account
for the increased workload required to achieve glycemic control
utilizing the tool. This allows us to plan for and justify the
additional nursing time required.
The manufacturer has been able to
provide interface with our application delivery system since implementation.
Additionally, our Information Technology department is working
closely with the manufacturer's engineers to develop an interface
with our electronic medical record. In the interim, we continue
to enter CBG values manually and utilize the software-generated
label for documentation of recommendations.
We've also implemented a “softer”
lancet, which presents less discomfort to patients from increased
fingersticks, and we've minimized using the central-line for blood
sampling for CBGs as to not increase the potential central-line-associated
bloodstream infections from frequent line interruptions.
Return on investment
Within our program, we're investigating the tool as a solution
in transitioning patient from N.P.O. or continuous enteral nutrition
status (and continuous insulin infusion) to intermittent oral
intake (and correction dose insulin). The tool is designed primarily
for use with steady carbohydrate intake. Although the tool can
derive a basal subcutaneous insulin regimen, including prandial
and correction dosing, for patients who are eating, we utilized
this only in a small group of patients. Utilization of this regimen
requires daily evaluation of insulin requirements as the patient's
resistance changes.
Because our trial focused specifically
on postoperative cardiovascular surgery patients, specific efficacy
with other ICU patient populations needs further study. We're
also unable to assess the tool's value in the step-down area.
Hyperglycemia is becoming increasingly
recognized as a significant factor in increased mortality particularly
in the postcardiac surgery patient. Our experience with standard
insulin nomograms and empirical attempts to customize glycemic
control hasn't been successful. Several software-based glycemic
control programs exist. During our trial with the software-based
tool, we observed a 67% improvement in control time, an 84% decrease
in hypoglycemia incidence, an 80% increase in CBGs within the
optimal range, and an 84% decrease in hyperglycemic values when
comparing prior values to the tool values. The projected cost
savings from decreased length of stay and mortality (such as prevention
of SSIs) further supports the cost and labor expense associated
with the software and increased testing.
References
1. Zerr KJ, Furnary AP, Grunkemeier GL, et al. Glucose control
lowers the risk of wound infection in diabetics after open-heart
operations. Ann Thorac Surg. 1997; 63(2): 356–361.
2. Furnary A, YingXing Wu, Bookin S. Effect of hyperglycemia and
continuous intravenous insulin infusions on outcomes of cardiac
surgical procedures: the Portland diabetic project. http://www.providence.org/resources/oregon/images/portlandprotocol/pdfs/EndocrPract04.pdf.
3. Latham R, Lancaster AD, Covington JF, et al. The association
of diabetes and glucose control with surgical-site infections
among cardiothoracic surgery patients. Infect Control Hosp
Epidemiol. 2001; 22(10): 607–612.
4. Van den Berghe G, Wouters PJ, Bouillon R, et al. Outcome benefit
of intensive insulin therapy in the critically ill: insulin dose
versus glycemic control. Crit Care Med. 2003;31(2):359–366.
5. Institute for Healthcare Improvement: Implement Effective Glucose
Control. http://www.ihi.org/IHI/Topics/CriticalCare/IntensiveCare/Changes/ImplementEffectiveGlucoseControl.htm.
6. AHA: www.aha.org.
Source: Nursing2009 Critical
Care. March 2009.
|