Evaluation of Point of Care Bedside
Glucose Monitors for Use in a Specialty &
Transplant Hospital
Mary A. Weidner, BS, MT(ASCP)
Independent Project HCAI 5221
Background
Methodist Specialty and Transplant Hospital is a 379 bed
facility providing a variety of organ transplant services
A Roche AccuCheck POC device is used post-operatively at
our facility to monitor whole blood glucose in the ICU.
A positive bias has been observed for the Roche AccuChek
whole blood glucose meter (Fig 1 is an example of this
observation)
A review of the medical literature revealed a number of
substances (e.g., maltose, hematocrit, IVIG) can interfere
with whole blood glucose meters.
This potential for interference and subsequent discordance
with central lab methods led to concern by the clinical staff
about accuracy of POCT results and exploration of
alternative glucose monitors
Objectives
Through regression analysis evaluate the
accuracy, specificity, connectivity, and
clinical performance of alternative POC
glucose monitoring devices against the
central laboratory for use within
Methodist Specialty and Transplant
Hospital
Materials and Methods
Prior to testing these devices in a clinical setting, in lab linearity and
precision studies were performed (data not shown). Each monitor tested
performed acceptably. To assess the potential for interference two
glucose meters available at the time (AccuChek and Precision XP) were
evaluated in patients typical of our hospital. Arterial whole blood from
central lines was taken at the times indicated in Fig 1 and in parallel ran
on each meter. In addition these specimens were spun down and the
plasma ran on the Vista hexokinase method for comparison. Mean % bias
was calculated relative to the Vista reference method. Fig 1 and 2 show
the results of this comparison. A Nova StatStrip was acquired at a later
time. Studies to evaluate the potential for known interferences such as
hematocrit and maltose were assessed by testing samples of 3 different
glucose concentrations in which the hematocrit and maltose
concentration was adjusted to varying levels (Fig 3). Finally whole blood
specimens were spiked with increasing concentrations of glucose and
then ran on each glucose monitor. These specimens were subsequently
spun down and plasma ran on the Vista hexokinase central laboratory
method. Regression analysis was performed to determine concordance
between the two methods (Fig 4). The Point of Care team also evaluated
connectivity options based on ability to support existing POC tests and
the ability to expand as new POC tests are implemented.
Figure 1. Increased Bias Observed with Roche AccuChek. Arterial whole
blood measurements were performed post-operatively as a function of time.
Panel A shows a mean positive bias of 35% for the Roche AccuChek in a patient
administered IVIG. Panel B also demonstrates a positive mean bias (16%) in a
patient which underwent liver transplant procedure.
Panel A
Panel B
Figure 2. Bias Also Observed with Abbott Precision XP Glucose Monitor. Panel A
shows a mean positive bias of 11% when the same specimens from the patient 1 below
were ran on the Abbott Precision XP glucose meter. Panel B shows a mean positive bias
of 3% when the specimens below for patient 2 are ran on the Abbott Precision XP.
Relative to the Roche AccuChek, the Abbott Precision XP had a mean bias ~3x less for
patient 1 and ~5x less for Patient 2.
Patient 1: IVIG Therapy
AccuCheck vs. Vista
Panel A
Patient 2: Liver Recipient
AccuCheck vs. Vista
Panel B
HCT INTERFERENCE
Figure 3. Effects of Hct and Maltose on Whole Blood Measurement of Glucose. Three
different glucose concentrations were measured under conditions in which the hematocrit
and maltose concentration was adjusted to varying levels. In both instances the Nova
StatStrip was not affected. However the AccuChek device was negatively affected by
increasing Hct and positively affected by increasing maltose concentrations
at all glucose levels studied.
Maltose Interference
StatStrip
StatStrip
Accuchek
Accuchek
250
110
100
90
80
70
60
22
45
Hematocrit %
61
Glucose (mg/dL)
Glucose (mg/dL)
120
200
150
100
50
0
100
200
Maltose (mg/dL)
HCT Interference cont.
HCT Interference
Maltose Interference
StatStrip
StatStrip
Accuchek
Accuchek
350
Glucose (mg/dL)
Glucose (mg/dL)
300
250
200
150
100
22
45
Hematocrit %
61
300
250
200
150
0
100
200
Maltose (mg/dL)
HCT Interference cont.
Maltose Interference
HCT Interference
StatStrip
StatStrip
Accuchek
Accuchek
600
Glucose (mg/dL)
Glucose (mg/dL)
500
450
400
350
300
500
400
300
250
0
22
45
Hematocrit %
100
61
Maltose (mg/dL)
200
GLUCOSE METERS VS.
CENTRAL LAB
Figure 4 a, b, c respectively. Glucose meters vs.
central lab. All three glucose meters tested
demonstrated strong correlation with the central lab
method with R2 values shown. Linearity was
somewhat equal between the Abbott Precision
Xceed Pro and the Nova StatStrip with a slope of
0.959 to 0.998, respectively. The Roche AcccuCheck
demonstrated the highest y-intercept at 20.848.
Figure 4a.
Figure 4b.
Figure 4c.
Conclusions
All three meters demonstrate good
correlation with central lab, however
both the Abbott Precision Xceed Pro
(R2 = 0.98567) and the Nova StatStrip
(R2 = 0.99790), had the acceptable
CLSI correlation with slopes closest to
1 and lower y-intercept then the Roche
AccuCheck meter.
Future Directions
•Future directions will include investigations to evaluate the
performance of the Abbott Precision XP in clinical settings.
The decision to go with the Abbott XP was made by the MHS
system because it was the best overall fit for the entire MHS
System and its patient population.
•Electronic Diabetes Flow Record will be utilized to perform
“real time” patient treatment, which currently All hospitals
have a lag time from result to actual intervention. With this
electronic record, once the glucometers go wireless; MHS
will be leading the nation in the innovation of patient
treatment so that it will no longer will only be about insulin
but inclusive of all diabetes treatment.