Glucoflex-R
Blood Glucose Test Strips
Evaluation in the hands of lay users - visually interpreted.
Authors:
D Mohr
A Bolek
D Mihaila
© National Diagnostic Products 2007
Contents
Introduction……………………..............................................................………….........…………….3
Description of Glucoflex-R test strips…………………..........…...…….............................................4
Comparative Method……………………………………………….....…............................................4
Study Outline…………………………………………………………..................................................5
Range of Results……………………………………………………....................................................5
Bias……………………………………………………………………..................................................6
Linear Regression…………………………………………………....................................................7
ISO 15197 breakdown of results……………………………………...................................................8
References……………………………………………………………...................................................9
Introduction
Diabetes is now a major public health problem. The number of people with diabetes is estimated
to reach 330 million by 20301. There is a high burden from the disease: people with diabetes have
an increased two to four fold risk of stroke and heart disease compared to the general population,
and increased incidence of retinopathy, peripheral nerve damage and renal problems. There is now
strong evidence for the effectiveness of tight glycemic control in reducing complications among
people with diabetes2.
National Diagnostic Products is the manufacturer of Glucoflex-R blood glucose test strips. The test
strip is designed to be read in a reflectance meter (Reflolux S) or visually by comparing the reacted
strip against a colour chart located on the side of the test strip container.
The purpose of this study was to evaluate the performance of Glucoflex-R when visually interpreted
by lay users. Previous studies have demonstrated the performance of the test strip when visually
interpreted by trained technicians and when used in combination with the Reflolux-S meter.
To provide proof of its measurement reliability and analytical performance in the hands of lay users,
the Glucoflex-R test strip was assessed against a well established laboratory method, Yellow Springs
Instruments 2300 STAT PLUS Analyser (YSI).
Description of Glucoflex-R Test Strips
Glucoflex-R is a dry reagent test strip based on colorimetric technology. It is designed for use with
capillary blood obtained from a finger stick. It is intended for self testing, accident and emergency
and ward use.
The test uses the glucose oxidase peroxidise reaction which along with the leuco dyes TMB, APAC
and o-Tolidine produces a colour change that is directly proportional to the concentration of glucose
contained in the blood. The test utilises the plasma fraction of the blood but is calibrated to report
results equivalent to whole blood. The measurement range is 1-44mmol/l when visually interpreted.
When used as a visual strip there is no calibration required. The test strips are packed in a plastic
bottle closed with a desiccant containing cap. The strip requires a blood volume of at least 20 μL.
Blood is applied to the test strip and wiped after one minute. After waiting a further minute, results
are read by comparing the strip to a colour chart on the label. Results greater than 17mmol/l are
read at three minutes. Times for reading are indicated on the side of the label. Colour photos are also
provided on the label depicting the test steps.
Designated Comparative Method
The comparative method used in this evaluation was the YSI 2300 Stat Plus Analyser. This system
determines the glucose concentration in a 25μL whole blood or plasma sample within 65 seconds.
The device has a broad measurement range of up to 900mg/dL and automatically calibrates ready
for use. It accurately measures glucose concentration with an overall error of prediction of 0.28mM
and a mean percent error of 1.79%. The YSI 2300 Stat Plus Analyser has a precision of ±2% of the
reading.
The system uses an enzyme immobilised in between two layers, which oxidises the substrate to
produce hydrogen peroxide. In turn the hydrogen peroxide is oxidised resulting in a current, which is
proportional to the substrate concentration. The system is traceable to an NIST standard.
Glucoflex-R Performance Evaluation
A lay-user performance evaluation for Glucoflex-R was conducted with 110 subjects, over a 5 week
period. Subjects were diabetic patients attending the collection rooms of St.Vincents Hospital,
Darlinghurst.
110-paired tests were performed by 110 lay users using Glucoflex-R strips and by trained technicians
using the YSI 2300, as the comparative method.
After Informed Consent was obtained, participants were provided with training in the use of
Glucoflex-R test strips. Each participant was given a reacted strip and shown how to match the
colour. The following instructions were given;
1. Hold the strip against the label and ask the question is the strip colour lighter or darker than the
label colour block. Move to the next colour block. Repeat until the nearest colour match is found.
2. If the test strip colour is between two colour blocks on the label then an average of the two values
should be calculated e.g.
For a strip that lies between colour blocks 9 and 11: (9+11)/2 = 10.
3. If participants judge the result to be some other value such as just under or over a colour block
then an approximation can be made. (it was found this was the most common method used rather
than strict calculation).
Subjects were then given an unopened Glucoflex-R pack and asked to follow the instructions and
perform a test unassisted.
The samples were fresh capillary blood, obtained directly from a finger-prick in all cases.
The glucose concentrations, as measured with the YSI 2300 system, span the intervals specified in
Table 1, ranging from 3.2mmol/L to 28.1mmol/L.
Table 1
Glucose concentrations of samples for system accuracy evaluation
Percentage of samples
(%)
15.5
20.0
27.3
23.6
9.1
4.6
(mmol/L)
2.8-4.3
4.4-6.7
6.8-11.1
11.2-16.6
16.7-22.2
>22.2
The plot of bias is presented in Figure 1. The graph shows the differences between Glucoflex-R and
YSI 2300 results plotted against the YSI 2300, with a pattern suggesting an increase of bias with
concentration.
There was a statistically significant overall average bias of 0.38mmol/L, against YSI 2300. Results
were on average less than 4% higher than those obtained using YSI 2300, with less than 15% of
results showing an absolute bias of more than 15%.
Figure 1
B i a s p lo t G lu c o f le x - R v s . Y S I 2 3 0 0
8 .0
6 .0
4 .0
2 .0
0 .0
- 2 .0
- 4 .0
Glucoflex-R - YSI 2300 (mmol/L)
- 6 .0
- 8 .0
0 .0
5 .0
1 0 .0
1 5 .0
Y S I 2 3 0 0 ( m m o l/L )
2 0 .0
2 5 .0
3 0 .0
Linear regression studies (ordinary linear regression) for Glucoflex-R compared with YSI 2300 are
presented in Figure 2, together with the error grid analysis.
Figure 2
Error grid analysis for blood glucose monitoring devices was developed (Clarke et al3) in an attempt
to evaluate the clinical performance of such devices used for self-monitoring. The analysis indicates if
the results obtained are clinically accurate and acceptable. For that purpose, the results are classified
in five zones as follows:
-
Zone A represents results which differ by <20% from the reference
-
Zone B represents results that differ from the reference by >20%, but would lead to “benign or no treatment”
-
Zone C represent results that would lead to “inappropriate intervention to alter an accept
able glucose concentration”
-
Results in Zone D express “ dangerous failure to detect and treat errors”
-
Results in zone E would lead to “ erroneous treatment”.
To be clinically acceptable, a blood glucose-monitoring device has to give 95% of the glucose
measurements in Zone A and no result in Zones C-E.
The results obtained with Glucoflex-R test strips showed good correlation with YSI 2300 method.
Zone A of the error grid contained more than 96% of results with all remaining results situated in
Zone B.
Glucoflex-R individual results are plotted against YSI 2300 results. The standard error of the
estimate was 1.03 mmol/L .The correlation coefficient of 0.99 (p<0.001) confirms a high correlation
of Glucoflex-R results with those obtained using YSI 2300.
The graph shows the regression line, the line of equivalence (45°line) that would be seen if the
Glucoflex-R and YSI 2300 gave identical results.
The slope and intercept with their standard errors and 95% confidence intervals are presented in
Table 2.
Table 2
Regression statistics of Glucoflex-R against YSI 2300
Intercept (mmol/L)
Slope
Estimate (standard error)
-0.053 (0.195)
1.043 (0.017)
95% confidence interval
-0.44 - 0.34
1.01 - 1.08
Results met ISO151974 acceptance criteria limits:
• at glucose concentrations <4.2mmol/l, 95% of the individual glucose results shall fall within
±0.83mmol/L of the results of the reference method
• at glucose concentrations ≥4.2mmol/L, 95% of the individual glucose results shall fall within ±
20% of the results of the reference method
Tables 3 and 4 present the accuracy of Glucoflex-R segregated for glucose levels <4.2mmol/L and
for glucose levels ≥4.2mmol/L. The number of samples tested, and the percentage results within
specified ranges are shown.
Table 3 Glucoflex-R accuracy for glucose concentrations <4.2mmol/L
(19 samples were tested in this range) Within±0.28mmol/L
Within±0.54mmol/L
Within±0.83mmol/L
8/19 (42%)
14/19 (74%)
19/19 (100%)
Table 4 Glucoflex-R accuracy for glucose concentrations ≥4.2mmol/L
(91 samples were tested in this range)
Within±5%
Within±10%
Within±15%
Within±20%
36/91 (40%)
62/91 (68%)
77/91 (85%)
87/91(96%)
19 samples were tested in the range of glucose <4.2mmol/L, with no result outside the ±0.83mmol/L
interval, relative to YSI 2300 results.
In the range of glucose ≥4.2mg/dL, 91 samples were tested with 15% of results outside the ±15%
interval and 4% outside the ±20% interval, relative to YSI 2300 results.
References
Wild S, Roglic G, Green A, Sicree R, King H: Global Prevalence of Diabetes: Estimates for the year
2000 and projections for 2030. Diabetes Care 2004, 27:1047-1053.
1
UK Prospective Diabetes Study (UKPDS) Group: Intensive blood glucose control with sulfonylurea
or insulin compared with conventional treatment and risk of complications in patients with type II
diabetes (UKPDS 33). Lancet 1998, 352,837-853.
2
Clarke WL, Cox D, Gonder-Frederick LA, Carter W, Pohl SL : Evaluating clinical accuracy of systems
for self-monitoring of blood glucose. Diabetes care 1987;10 (5): 622-628.
3
ISO 15197:2003 In vitro diagnostic test systems - Requirements for blood glucose monitoring
systems for self-testing in managing diabetes mellitus.
4