a 5-center crc-based study of the accuracy of the glucowatch® g2 tm biographer in children and...
TRANSCRIPT
A 5-center CRC-based Study of the Accuracy of the GlucoWatch® G2TM
Biographer in Children and Adolescents with Type 1 Diabetes
Darrell Wilson1, Bruce Buckingham1, Roy Beck2, Katrina Ruedy2, Craig Kollman2, Eva Tsalikian4, Timothy Wysocki5, Stuart Weinzimer3, Peter Chase6 and the Diabetes Research in Children Network (DirecNet) Study Group 1Stanford, CA; 2Tampa, FL; 3New Haven, CT; 4 Iowa City, IA; 5Jacksonville, FL; 6Denver, CO.
ABSTRACT A 5-center CRC-based Study of the Accuracy of the GlucoWatch® G2TM Biographer in Children and Adolescents with Type 1 Diabetes. Darrell Wilson1, Bruce Buckingham1, Roy Beck2, Katrina Ruedy2, Craig Kollman2, Eva Tsalikian4, Timothy Wysocki5, Stuart Weinzimer3, Peter Chase6 and the Diabetes Research in Children Network (DirecNet) Study Group 1Stanford, CA;2Tampa, FL; 3 New Haven, CT; 4 Iowa City, IA; 5Jacksonville, FL; 6Denver, CO. An accurate glucose sensor would greatly improve the management of diabetes in children, particularly at night. We conducted a study of 87 children with T1DM to assess the accuracy of the GlucoWatch G2 Biographer (“GWB”). While subjects wore 1 or 2 GWBs, we obtained gold standard venous glucose concentrations (GS) at 30-60 min intervals over a 24hr period and every 5 min during a meal-induced hyperglycemia test and an insulin-induced hypoglycemia test. The 87 subjects included in the analysis ranged in age from 3 to 17 (mean 10.0) years; 49% were female and 87% Caucasians. Among the 14,777 possible sensor glucose measurements, 2,586(18%) were skipped. Accuracy according to blood glucose level is indicated below.
Gold Standard Blood Glucose Level (mg/dl)
<70 71-120 121-180 181-240 >240 Total
# of paired data points 334 926 955 766 654 3635
Difference (GWB-GS) mg/dl median 22 9 -1 -6 -10 3
Relative absolute difference* median (absolute value of GWB-GS)/GS
0.38 0.18 0.14 0.14 0.12 0.16
% of values within 20% 28% 55% 66% 67% 67% 60%
For the 3635 paired values, the Pearson correlation was 0.86. Detection of GS values <70 mg/dl had a sensitivity of 35%, specificity of 97%, positive predictive value of 57%, and negative predictive value of 94%. Accuracy showed no meaningful variation by subject age. In terms of precision, the sensor values from two simultaneously-worn GWBs had a correlation of 0.85. Skin reactions were generally mild; there were no severe reactions. GWB glucose values were more accurate for hyperglycemic than hypoglycemic glucose levels. In summary, while the GWB is able to track glucose trends and is accurate within 20% of the blood glucose the majority of the time, its utility as a management tool for children with diabetes requires further study in a longitudinal outpatient study. Funded by the following NIH grants:HD041890, HD041915, HD041918-01, HD041908-01, HD041919-01, and HD041906-01.
Background
• The Diabetes Research in Children Network (DirecNet) is a NIH-funded collaborative study group.
• The major objective of DirecNet is to critically evaluate the clinical usefulness of current and future glucose sensor devices in youth with T1DM.
Goals
• The purpose of this study was to determine the accuracy of the GlucoWatch® G2™ Biographer (GW2B) in children with Type 1 diabetes mellitus (T1DM).
GlucoWatch® G2™ Biographer (GW2B)
• Measures interstitial glucose extracted by reverse iontophoresis
• Requires a single calibration value following the 2-hour warm-up period
• Glucose readings every 10 minutes for 13 hours
• Values range from 40 to 400 mg/dL
Subject Selection
• Age 1 to 18 years
• T1DM for at least 1 year
• BMI between 5th and 95th percentile for age and gender
• Weight ≥ 12.0 kg if < 7 yrs and ≥ 16.0 kg if ≥ 7 yrs
• Normal hematocrit
• No skin or other significant disorders
Study Procedures
• Admitted to Clinical Research Center• Approximately 24 hours
• Two GW2Bs placed at staggered times• Sensor pads were replaced as needed
• Venous blood sampling• Hourly during day (7:00AM to 9:00PM)• Twice hourly during night (9:30PM to
6:30AM)• Central lab for reference glucoses
Study Procedures
• Meal induced hyperglycemia• high carbohydrate drink
• 1.75 gm CHO/kg, maximum of 75 grams
• Samples q 5 minutes for 60 minutes
• Insulin induced hypoglycemia• Subjects older than 7 years only• 0.05-0.10 units/kg regular insulin IV bolus• After 30 minutes, a second dose if needed
• Glycemia goal of ≤ 55 mg/dL
• Samples q 5 minutes for up to 90 minutes
GlucoWatch Function• Sensor calibration
• 1st attempt 94%• 2nd attempt 2%• Failed 4%
• Sensor lifespan*• ≥14.5 hours (Full) 81%• 10.0 to 14.5 hours 3%• 6.0 to 10.0 hours 6%• <6.0 hours 10%
• Of 14,232 possible GW2B measurements• 17% were skipped
* Excludes sensors stopped early for logistical reasons.
Evaluation of Accuracy
The appropriate set of accuracy measures to evaluate near continuous glucose monitoring remains to be developed. We used the traditional methods that are utilized to assess the accuracy of blood glucose meters. These methods do not adequately capture the time dimension of glucose sensor data with respect to glucose trend (slope) information or glucose pattern detection across the day.
Measures of Accuracy
• Pearson’s correlation• Difference
• sensor – reference
• Relative difference %• (sensor – reference)/reference x 100%
• Absolute difference• |sensor – reference|
• Relative absolute difference (RAD) %• |(sensor – reference)/reference| x 100%
Measures of Accuracy
• ISO criteria (proposed)• Reference glucose ≤75 mg/dL
• Sensor value within ±15 mg/dL
• Reference glucose value >75 mg/dL• Sensor value within ±20%
• Error grid analysis• Clarke• Consensus
Subject Characteristics89 Subjects
Gender 44 (49%) F 45 (51%) M
Age (years) mean ± SD 9.9 ± 4.1
Duration of diabetes (years) mean ± SD
4.6 ± 3.2
Insulin delivery39 (44%) pump
50 (56%) injections
HbA1c mean ± SD 7.8% ± 1.2%
BMI percentile mean ± SD 67% ± 23%
A Subject
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
3:00PM
4:00PM
5:00PM
6:00PM
7:00PM
8:00PM
9:00PM
10:00PM
11:00PM
12:00AM
1:00AM
2:00AM
3:00AM
4:00AM
5:00AM
6:00AM
7:00AM
8:00AM
9:00AM
10:00AM
11:00AM
12:00PM
1:00PM
2:00PM
3:00PM
Glu
co
se
(m
g/d
L)
Reference GW2B
GW2B Accuracy Summary3,672 paired values
• Median difference, 3 mg/dL,• Not significantly different than zero
• Median absolute relative difference• 16%
• Correlation• 0.86
• Consensus Error Grid• 97%, A+B zones
GW2B Accuracy
Mean
(95% CI)
Median
(25th to 75th%-ile)
Differencemg/dL
2.2(-2.0, 6.1)
3(-19, 26)
AbsoluteDifference mg/dL
30.9(28.7, 33.2)
23(11, 42)
Relative
Difference6%
(4%, 9%)
2%(-12%, 20%)
Relative Absolute
Difference22%
(20%, 23%)
16%(7%, 29%)
Factors NOT Impacting Accuracy
• Age
• Gender
• BMI (body mass index)
• Time (6:30AM -10:30PM vs 11PM – 6AM)
• Location of GWB placement• Upper vs lower arm• Inner vs outer arm
Factors Impacting AccuracySensor Age
Sensor
age (hrs)
Number of paired values
RAD(median)
ISO criteria
met (%)
0 to <6 hr 1,316 15% 63%
6 to <10 hr 1,142 15% 64%
≥10 hr 1,214 19% 54%
p value 0.002 0.006
Factors Impacting AccuracyGlucose Concentration
Glucose
(mg/dL)
Number of paired values
RAD(median)
ISO criteria
met (%)
≤ 70 334 38% 32%
71-120 926 18% 55%
121-180 963 14% 66%
181-240 776 14% 66%
>240 673 13% 67%
p value <0.001 <0.001
A
A
B
C
D
E
B
C
D
E
0
50
100
150
200
250
300
350
400
450
500
550
0 50 100 150 200 250 300 350 400 450 500 550
Reference Blood Glucose mg/dL
GW
2B m
g/d
LClarke Error Grid
A
A
B
CD
B
C
D
E
0
50
100
150
200
250
300
350
400
450
500
550
0 50 100 150 200 250 300 350 400 450 500 550
Reference Blood Glucose mg/dL
GW
2B m
g/d
L
Consensus Error Grid
Detection of Hypoglycemia
• Episode defined as at least two GW2B values ≤ 60 mg/dL without any values > 70 mg/dL.
• Distinct episodes had to be separated by at least 30 minutes.
• Episode was considered confirmed if there was at least one reference glucose value ≤ 70 mg/dL.
• Only 10 of 18 hypoglycemic episodes detected by the GW2B overnight were confirmed.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
1-Specificity
Sen
siti
vity
Alarm=60
Alarm=100
Alarm=80
Alarm=120
Receiver Operator Curve (ROC) for Hypoglycemia ( ≤60 mg/dL)
Alarm
Setting
Sensitivity False Alarm
Rate
60 23% 51%
80 59% 67%
100 84% 80%
120 92% 85%
Side Effects
• No serious skin reactions
• No other problems
Moderate Irritationexample
Mild Irritationexample
SummaryIn summary, our results are very similar to those reported in studies conducted by Cygnus, Inc. We found the GW2B to be more accurate during periods of hyperglycemia than hypoglycemia. The accuracy of this generation of the sensor does not approach the accuracy of current home glucose meters. However, the sensor glucose values are generally within 20% of the serum glucose. This level of accuracy may be sufficient for detecting trends and modifying diabetes management.
Acknowledgements
• This study would not have been possible without the support of CRC Nurses at the study sites.
• This research has been supported by the following NIH/NICHD Grants: HD041919-0; HD041915; HD041890; HD041918-01; HD041908-01; and HD041906-01.
• Clinical Centers also received funding through the following GCRC Grant Numbers M01 RR00069; RR00059; M01 RR 06022 and 5M01RR00070-41 and Nemours Biomedical Research.
• University of Minnesota Central Laboratory• Michael Steffes, Jean Bucksa, Maren Nowicki