958754

research-article2020
DSTXXX10.1177/1932296820958754Journal of Diabetes Science and TechnologyAlva et al

Original Article

Journal of Diabetes Science and Technology

Accuracy of a 14-Day Factory-Calibrated

2022, Vol. 16(1) 70­–77
© 2020 Diabetes Technology Society
Article reuse guidelines:
Continuous Glucose Monitoring System sagepub.com/journals-permissions
DOI: 10.1177/1932296820958754
https://doi.org/10.1177/1932296820958754

With Advanced Algorithm in Pediatric journals.sagepub.com/home/dst

and Adult Population With Diabetes

Shridhara Alva, PhD1 , Timothy Bailey, MD2, Ronald Brazg, MD3,

Erwin S. Budiman, PhD1, Kristin Castorino, DO4,
Mark P. Christiansen, MD5, Gregory Forlenza, MD6,
Mark Kipnes, MD7, David R. Liljenquist, MD8, and Hanqing Liu, MA1

Abstract
Background: In this study, we evaluated the analytical performance of the second-generation factory-calibrated FreeStyle
Libre Flash Glucose Monitoring (FreeStyle Libre 2) System compared to plasma venous blood glucose reference, Yellow
Springs Instrument 2300 (YSI).
Methods: The study enrolled participants aged four and above with type 1 or type 2 diabetes at seven sites in the United
States. Adult participants (18+ years) participated in three in-clinic sessions and pediatric participants (4-17 years) participated
in up to two in-clinic sessions stratified to provide data for days 1, 2, 3, 7, 8, 9, 12, 13, or 14 of sensor wear. Participants aged
11+ underwent supervised glycemic manipulation during in-clinic sessions to achieve glucose levels across the measurement
range of the System. Performance evaluation included accuracy measures such as the proportion of continuous glucose
monitoring (CGM) values that were within ±20% or ±20 mg/dL of reference glucose values, and bias measures such as the
mean absolute relative difference (MARD) between CGM and reference values.
Results: Data from the 144 adults and 129 pediatric participants were analyzed. Percent of sensor results within ±20%/20 mg/
dL of YSI reference were 93.2% and 92.1%, and MARD was 9.2% and 9.7% for the adults and pediatric participants, respectively.
The System performed well in the hypoglycemic range, with 94.3% of the results for the adult population and 96.1% of the
data for pediatric population being within 15 mg/dL of the YSI reference. The time lag was 2.4 ± 4.6 minutes for adults and
2.1 ± 5.0 minutes for pediatrics.
Conclusions: The System demonstrated improved analytical accuracy performance across the dynamic range during the
14-day sensor wear period as compared to the previous-generation device.
NCT#: NCT03607448 and NCT03820050

Keywords
FreeStyle Libre 2, continuous glucose monitoring, optional alarm, factory calibration

Introduction
The Abbott FreeStyle Libre Flash Glucose Monitoring 1
Abbott Diabetes Care, Alameda, CA, USA
System (Abbott Diabetes Care, Alameda, CA), introduced in 2
AMCR Institute, Escondido, CA, USA
2014, was the first factory-calibrated continuous glucose 3
Rainier Clinical Research Center, Renton, WA, USA
monitoring (CGM) device available to people with both type 4
Sansum Diabetes Research Institute, Santa Barbara, CA, USA
5
1 diabetes (T1D) and type 2 diabetes (T2D).1-3 Factory- Diablo Clinical Research, Walnut Creek, CA, USA
6
calibrated CGM devices offer distinct advantages over the Barbara Davis Center for Diabetes, University of Colorado Denver,
Aurora, CO, USA
previous user-calibrated devices, including reduced burden, 7
DGD Clinic, San Antonio, TX, USA
reduction in test-strip usage, and removal of inaccuracy 8
Rocky Mountain Diabetes Center, Idaho Falls, ID, USA
imparted from user glucose testing.4-6 The improvement in
Corresponding Author:
device accuracy has also resulted in approval for insulin dos- Shridhara Alva, PhD, Abbott Diabetes Care, 1360 South Loop Road,
ing from CGM values (non-adjunctive use), further reducing Alameda, CA 94502, USA.
user burden.7,8 Direct dosing from CGM has since been Email: [email protected]
Alva et al 71

demonstrated to be safe and effective in several randomized that could place the subject at risk by glucose manipulation.
clinical trials.9-12 The expansion of CGM utility and reduc- No exclusions were made based on the type of medications
tion in technology burden has contributed to an increased or supplements.
adoption of CGM by people with both T1D and T2D.13-17 To ensure that the CGM data adequately covered the
This first-generation product is used to guide diabetes System’s entire reportable range (40-500 mg/dL), glucose
management decisions, including insulin dosing. The Free- levels were manipulated to induce high or low blood glu-
Style Libre 2 System (hereafter referred to as “System”) addi- cose levels through carbohydrate consumption and insulin
tionally provides optional hypoglycemia and hyperglycemia timing during the clinic sessions for participants aged 11
alarms. When configured, the device sends data to the receiver and above.
every minute to generate an alarm. This product secured the Adult participants were scheduled for up to three in-clinic
Conformitè Europëenne (CE) mark in 2018.18 The product sessions of ten hours on days 1-3, 7-9, and 12-14. Pediatric
was updated in 2020 with a new algorithm, based on the anal- participants ages 6-17 years were scheduled for up to two in-
ysis of additional reference paired sensor data, to further clinic sessions of eight hours on days 1-2, 7-9, or 12-14
improve accuracy. depending on their body weight. All sessions included com-
The study described here evaluated the accuracy of this parison of sensor readings to plasma venous glucose concen-
System with the new glucose algorithm. There was no change trations using a laboratory reference method (Yellow Springs
to the sensor or insertion mechanism compared to the on- Instrument YSI 2300 [hereafter “YSI]; YSI, Inc., Yellow
market device. The study enrolled participants aged four and Springs, OH). Participants had venous blood drawn every
above. Participants older than 11 years underwent glycemic 5-15 minutes for the duration of each clinic session for YSI
challenges to obtain sufficient data for the evaluation of sen- measurement. The samples were centrifuged within 15 min-
sor performance at low and high glucose concentrations. utes of the blood draw and tested on YSI within 15 minutes
afterwards. A heating pad was applied around the intrave-
nous site to keep the insertion site open. The CGM devices
Methods were masked to participants and clinic staff for the entire
sensor wear.
Study Design Participants or caregivers/study staff scanned both sen-
The System was evaluated in two prospective multicenter sors with the paired reader immediately following reference
studies enrolling 153 adults of ages 18 and above (at five sample collection. The first inserted sensor with the YSI-
clinical sites) and 144 pediatric participants of ages 4-17 (at sensor pair was used for the analysis of the System perfor-
four clinical sites) with T1D or T2D. Seven adult participants mance, while the second applied sensor was used for the
and five pediatric participants screen-failed or discontinued assessment of precision. Sensor values were paired with YSI
prior to the sensor insertion, and thus were not evaluable for values by choosing the sensor value that was closest in time
the study effectiveness assessment. The studies were regis- to the corresponding YSI blood draw, but no more than five
tered with clinicaltrials.gov (NCT# NCT03607448 and minutes before or after the YSI blood draw. Each blood draw
NCT03820050). sample was assayed on the YSI in duplicate and the average
All participants were required to wear the sensor for up to of the readings was used.
14 days. Participants aged 4-17 years were required to per-
form four self-monitoring of blood glucose (SMBG) mea-
surements per day, while participants above 18 years were
Device Description
required to perform eight SMBG measurements. All SMBG The FreeStyle Libre family of products is unique among
measurements were made using the built-in blood glucose existing interstitial glucose monitoring technologies that
meter that uses FreeStyle Neo test strips. Participants received uses a wired enzyme technology with glucose oxidase. The
no training on the devices (but were trained on study proce- sensor is “factory only” calibrated and has a wear time of up
dures) and inserted one sensor on the back of each of their to 14 days without any user calibration. This System has the
upper arms following the instructions for use. For pediatric same form factor (shape and size) as the first-generation
participants, either the parent or the participant inserted the product.
sensor. Sensors that failed within the first hour after insertion The user is required to scan the sensor to get the current
were replaced. glucose, glucose trend, and historic glucose up to eight
The participants aged 11 and above were required to have hours. It also has optional hypoglycemic and hyperglyce-
intensive insulin therapy (either multiple daily injections or mic alarms and when configured, the Bluetooth-enabled
continuous subcutaneous insulin infusion [CSII]) with sensor will send the glucose information to the receiver
known insulin sensitivity factor and were required to undergo (either the reader or smartphone running the FreeStyle
controlled manipulation of glucose levels during at least one LibreLink app) every minute to generate the alarm. The
of the in-clinic sessions. Exclusion criteria included preg- glucose algorithm has been updated to improve accuracy at
nancy, anemia, or any condition, per investigator discretion, the low end of the dynamic range.
72 Journal of Diabetes Science and Technology 16(1)

Data Analysis evaluable. Two adult and ten pediatric (including eight
aged between four and five years who did not have YSI
Only those matched YSI-sensor pairs where the sensor reference measurements) participants did not have paired
results were within the reportable range (40-500 mg/dL) sensor to YSI data; thus, they were excluded from the
were used for the performance evaluation, as shown in the accuracy assessments presented in Tables 3–7. This
tables 3–7. The agreement levels were calculated relative to resulted in 144 adult and 129 pediatric participants with
a glucose concentration difference (in mg/dL) when the ref- YSI reference measurements.
erence glucose value was up to 80 mg/dL and relative to a The overall performance of the System as per age group
normalized concentration difference (in %) otherwise, and is presented in Table 3. There were a total of 18 926
were evaluated at four different ranges: ±15%/±15 mg/dL, matched data pairs in the adult group and 6584 matched
±20%/±20 mg/dL, ±30%/±30 mg/dL, and ±40%/±40 mg/ data pairs in the pediatric group. The MARD of the adult
dL. The mean absolute relative difference (MARD) was cal- group was 9.2% (95% CI [8.7-9.9]) and of the pediatric
culated as the absolute value of the average percent differ- group was 9.7% (95% CI [8.9-10.7]). (For the four- to
ence between the paired sensor and reference glucose values. five-year-old participants, the MARD against SMBG ref-
The consensus error grid analysis was performed for paired erence was 11.8%. For comparison, MARD for the adult
glucose data.19 population against SMBG was 10.7%.) The System showed
Paired YSI-sensor data were grouped into four wear that 93.2% and 92.1% of sensor readings were within 20%
periods: beginning (days 1, 2, and 3 for adults, and days 1 or 20 mg/dL of YSI reference for the adult and pediatric
and 2 for pediatric participants), early middle (days 7 and groups, respectively. The subgroup analysis did not show
8), late middle (days 9 and 12), and end (days 13 and 14). any marked difference in overall performance relative to
The total drift of the sensitivity over wear period was age (P value = .6445), type of diabetes (P value = .2689),
assessed by performing a linear regression on the paired clinical site (P value = .3473), insulin administration
readings of relative difference between the sensor and (P value = .4216), or HbA1c (P value = .3554). The consen-
SMBG against sensor elapsed time. The between-sensor sus error grid analysis19 showed 93.2% and 92.6% of the
precision was calculated as the coefficient of variation data in the A zone and 99.9% and 100% of the data in the
(CV) from the paired glucose readings from two sensors A+B zone of the error grid for the adult and pediatric
worn simultaneously with matched wear time. groups, respectively.
The ability of the System to alarm appropriately when Relative agreement at different YSI glucose ranges is pre-
alerts were set at different thresholds was assessed by com- sented in Table 4. Both the adult and pediatric studies exhibit
paring sensor results to YSI measurements within a 15-min- similar agreement level across glucose ranges.
ute time window at low (60, 70, 80, and 90 mg/dL) and high The percentage within 20% or 20 mg/dL of YSI reference
(120, 140, 180, 200, 220, 240, and 300 mg/dL) glucose and MARD performance throughout the wear duration was
threshold levels and determining whether an alert would assessed for both adult and pediatric studies. The percentage
have been generated. At each threshold, true alarm rate within 20% or 20 mg/dL of YSI reference for both study pop-
(whether YSI is in agreement with the device when the ulations (Table 5) remains consistent across the wear peri-
device alerts) and detection rate (whether the device alerts ods. Percentages of the data within 20% or 20 mg/dL of YSI
when YSI is within the threshold) were calculated. The false reference for the adult and pediatric data on the first day of
alarm rate and missed detection rates were calculated as the sensor wear were 87.9% and 85.0%, respectively. The
(100% ─ true alarm rate) and (100% ─ detection rate) mean absolute difference (MAD) and MARD for both study
respectively. populations (Table 6) remained consistently low across the
All analyses were conducted using SAS 9.4 Software. wear periods. The MARD on the first day of the sensor wear
for the adult and pediatric data was 11.2% and 12.1%,
respectively.
Results
The total drift estimated over the 14-day wear duration
Demographics and baseline characteristics of the evaluable was 0.6% and -0.2% for the adult and pediatric populations,
participants in the study are provided in Tables 1 and 2. The respectively. Estimated probability of sensor surviving
subject population had a good spread of HbA1c values cov- 14 days was 71.1% for adults and 78.1% for the pediatric
ering a wide range and represented both CSII and multiple population. Of these, 4.1% for adults and 4.3% for pediatric
daily injection (MDI) users. The body mass index (BMI) population were due to the System terminating the sensor
ranged from 18.2 to 68.7 kg/m2 for adult population and from early, and the remaining failures were due to sensors getting
13.7 to 39.6 kg/m2 for the pediatric population. Greater than knocked off during wear.
90% of the participants had T1D. The alarm performance when alarms were set at different
Only data compared to the YSI reference are used for thresholds at low (60, 70, 80, and 90 mg/dL) and high (120,
evaluating the performance of the System. One hundred 140, 180, 200, 220, 240, and 300 mg/dL) glucose threshold
and forty-six adults and 139 pediatric participants were levels is presented in Table 7.
Alva et al 73

Table 1. Demographics of Study Subjects in Adult and Pediatric Populations.

Adult population Pediatric population

Demographic N % N %
Sex Female 78 53.4 62 44.6
Male 68 46.6 77 55.4
Race White—not Hispanic or Latino 127 87.0 109 78.4
White—Hispanic or Latino 11 7.5 26 18.7
Asian 4 2.7 0 0.0
Other 2 1.4 2 1.4
Black or African American 1 0.7 2 1.4
Native Hawaiian or Pacific Islander 1 0.7 0 0.0
Type of diabetes Type 1 133 91.1 137 98.6
Type 2 13 8.9 2 1.4
Insulin pump use Yes 88 60.3 101 72.7
No 58 39.7 38 27.3
HbA1c HbA1c < 7% 41 28.1 19 13.7
7% ≤ HbA1c ≤ 8.5% 74 50.7 64 46.0
HbA1c > 8.5% 31 21.2 56 40.3

Table 2. Characteristics of Study Subjects in Adult and Pediatric Populations.

Adult population Pediatric population

Characteristic Mean ± SD Range Mean ± SD Range

Age (years) 47.3 ± 15.9 18-82 12.2 ± 3.6 4-17
Weight Pounds 182.7 ± 47.1 106.2-438.6 115.4 ± 44.9 37.8-245.4
Kilograms 82.9 ± 21.4 48.2-198.9 52.3 ± 20.4 17.1-111.3
Height Inches 67.3 ± 4.1 60-81 60.7 ± 8.0 38-79
Meters 1.71 ± 0.10 1.52-2.06 1.54 ± 0.20 0.97-2.01
Body mass index (BMI) 28.3 ± 6.6 18.2-68.7 21.0 ± 4.6 13.7-39.6
Duration of diabetes (years) 26.0 ± 15.2 1.0-66.1 5.0 ± 3.2 0.2-16.0
Duration of insulin use (years) 24.8 ± 15.5 1.0-66.0 4.9 ± 3.2 0.2-15.0
Total number of injections per day 4.7 ± 1.5 1-9 5.1 ± 1.2 3-7
(N subjects = 38)
HbA1c (%) 7.8 ± 1.4 5.5-14.3 8.3 ± 1.4 5.6-13.7

Table 3. Overall Performance of the System by Subject Age Group.

Matched % within % within % within % within MARD,

Subject group Subjects pairs (n) ±15%/±15 mg/dL ±20%/±20 mg/dL ±30%/±30 mg/dL ±40%/±40 mg/dL %
Adults (ages 18+) 144 18 926 86.3 93.2 98.3 99.6 9.2
Pediatric (ages 6-17) 129 6584 84.5 92.1 97.2 99.0 9.7

Abbreviation: MARD, mean absolute relative difference.

The time lags between the venous reference and the sensor manage their diabetes.20 The System with improved accu-
result for adults and pediatric population were 2.4 ± 4.6 min- racy, specifically at the hypoglycemic ranges, also adds
utes and 2.1 ± 5.0 minutes, respectively. The mean CVs optional hypoglycemia and hyperglycemia alarms. The new
between the sensors in adult and pediatric population were System has the same form factor as the previous generation,
5.7% and 5.8%, respectively. and therefore the ease of use and comfort to wear have not
changed. The System is factory-calibrated without the need
for fingerstick calibration by the user.
Discussion Studies on the System with the new glucose algorithm
The FreeStyle Libre has allowed easy access to the sensing have demonstrated a performance of 93.2% and 92.1% of
technology with over two million patients using the device to the results within ±20% or ±20 mg/dL of YSI reference for
74 Journal of Diabetes Science and Technology 16(1)

Table 4. Accuracy Performance at Different Reference Glucose Levels (YSIref): (a) Data From the Adult Population; (b) Data From the
Pediatric Population; and (c) Data From the Adult and the Pediatric Populations in the Medically Relevant Glucose Ranges.
(a)
YSIref glucose level, mg/dL % within % within % within % within MAD, mg/dL/
[mmol/L]b ±15%/±15 mg/dL ±20%/±20 mg/dL ±30%/±30 mg/dL ±40%/±40 mg/dL MARD, %a N

40-50 [2.2-2.8] 87.4 96.3 99.5 100.0 9.1 215

51-80 [2.8-4.4] 91.2 97.3 99.7 100.0 7.0 4483
81-180 [4.5-10.0] 78.7 87.9 96.1 99.3 10.1 6319
181-300 [10.0-16.7] 89.7 94.8 99.0 99.6 7.5 5311
301-400 [16.7-22.2] 90.0 95.7 99.3 99.6 7.1 2492
401-500 [22.3-27.8] 78.3 89.6 99.1 100.0 10.2 106
Overall 86.3 93.2 98.3 99.6 9.2 18 926

a
Mean absolute difference (MAD) is provided for glucose levels ≤80 mg/dL, and mean absolute relative difference (MARD) is provided for glucose levels >80 mg/dL.
b
Accuracy results for glucose values ≤80 mg/dL are in mg/dL.

(b)
YSIref glucose level, % within % within % within % within MAD, mg/dL/
mg/dL [mmol/L]b ±15%/±15 mg/dL ±20%/±20 mg/dL ±30%/±30 mg/dL ±40%/±40 mg/dL MARD, %a N

40-50 [2.2-2.8] 88.7 96.8 100.0 100.0 8.8 62

51-80 [2.8-4.4] 92.3 96.6 99.5 100.0 6.8 1213
81-180 [4.5-10.0] 75.0 85.1 94.0 97.7 11.3 2350
181-300 [10.0-16.7] 87.6 95.0 98.3 99.3 8.3 2058
301-400 [16.7-22.2] 91.3 97.2 99.4 100.0 7.3 871
401-500 [22.3-27.8] 96.7 100.0 100.0 100.0 6.0 30
Overall 84.5 92.1 97.2 99.0 9.7 6584

Mean absolute difference (MAD) is provided for glucose levels ≤80 mg/dL, and mean absolute relative difference (MARD) is provided for glucose levels >80 mg/dL.
a

Accuracy results for glucose values ≤80 mg/dL are in mg/dL.

b

(c)
Adult population Pediatric population

YSIref glucose level, No. % within % within % within No. % within % within % within
(mg/dL) [mmol/L]a pair ±15%/±15 mg/dL ±20%/±20 mg/dL ±40%/±40 mg/dL pair ±15%/±15 mg/dL ±20%/±20 mg/dL ±40%/±40 mg/dL

<70 [<3.9] 3473 94.3 98.4 100.0 882 96.1 98.8 100.0
70-180 [3.9-10.0] 7544 77.3 86.8 99.4 2743 74.3 84.8 98.0
>180 [>10.0] 7909 89.6 95.0 99.6 2959 88.7 95.7 99.5

Accuracy results for glucose values <70 mg/dL are in mg/dL.

a

Table 5. Percentages Within 20%/20 mg/dL of YSI reference Performance at Different Wear Periods for Adult and Pediatric
Populations.
Adult population Pediatric population

Wear period (days) Within ±20%/±20 mg/dL Outside ±40%/±40 mg/dL N Within ±20%/±20 mg/dL Outside ±40%/±40 mg/dL N

Beginning (1-3) 91.2 0.7 7113 89.1 1.5 1841

(Day 1)a (87.9) (1.6) (2366) (85.0) (2.0) (1090)
Early middle (7-8) 95.1 0.2 4541 94.5 1.5 1651
Late middle (9-12) 94.2 0.3 3512 93.3 0.5 1542
End (13-14) 93.7 0.0 3760 91.9 0.6 1550
Overall 93.2 0.4 18 926 99.0 1.0 6584

a
Performance on Day 1 is provided in parentheses with the data for the beginning phase of the sensor wear.

adult and pediatric populations, respectively, with over 25 population being within 15 mg/dL of the YSI reference.
000 paired datapoints between the two studies—a perfor- Accuracy of the sensor is stable over the sensor wear period
mance that is comparable to that reported for the Dexcom and the performance on the first day of sensor wear (with
G6 system (92.3%).21,22 The System performed well in the the results within 20% or 20 mg/dL of the YSI reference
hypoglycemic range with 94.3% of the results for the adult was 87.9% for adults and 85.0% for pediatric populations)
population and 96.1% of the results for the pediatric is comparable to the reported numbers for the G6 system
Alva et al 75

Table 6. MARD Performance at Different Wear Periods for Adult and Pediatric Populations.

Adult population Pediatric population

Wear period Mean, % Median, % N Mean, % Median, % N

Beginning (1-3) 10.0 7.6 7113 10.7 8.3 1841
(Day 1)a (11.2) (8.2) (2366) (12.1) (9.6) (1090)
Early middle (7-8) 8.5 6.8 4541 8.0 5.6 1651
Late middle (9-12) 8.8 6.7 3512 9.7 8.0 1542
End (13-14) 9.1 6.8 3760 10.2 8.7 1550
Overall 9.2 7.1 18 926 9.7 7.6 6584

Abbreviation: MARD, mean absolute relative difference.

a
Performance on Day 1 is provided in parentheses with the data for the beginning phase of the sensor wear.

Table 7. Glycemic Alarms From the Adult and Pediatric Populations.

Adult population Pediatric population

Threshold, mg/dL [mmol/L] Detection rate, %/N True alarm rate, %/N Detection rate, %/N True alarm rate, %/N
Hypoglycemic range 60 mg/dL 75.7/1527 72.3/9926 87.4/373 62.9/2780
[3.3 mmol/L]
70 mg/dL 89.3/3655 86.0/21 610 93.5/963 80.3/6363
[3.9 mmol/L]
80 mg/dL 97.3/4760 91.3/32 904 96.4/1318 85.6/9747
[4.4 mmol/L]
90 mg/dL 98.5/5600 93.6/41 447 97.3/1656 92.2/12 550
[5.0 mmol/L]
Hyperglycemic range 120 mg/dL 98.2/11 472 99.1/105 958 98.2/4441 98.8/34 176
[6.7 mmol/L]
140 mg/dL 98.1/10 203 99.1/93 918 98.4/3945 98.0/30 107
[7.8 mmol/L]
180 mg/dL 97.8/8113 99.2/74 492 98.0/3125 98.4/22 430
[10.0 mmol/L]
200 mg/dL 97.0/7296 99.2/66 214 98.0/2791 98.0/19 425
[11.1 mmol/L]
220 mg/dL 96.9/6413 99.0/57 710 96.9/2492 98.2/16 371
[12.2 mmol/L]
240 mg/dL 95.6/5570 98.4/48 894 95.7/2172 98.0/13 359
[13.3 mmol/L]
300 mg/dL 90.0/2688 96.3/21 629 91.0/962 90.8/6064
[16.7 mmol/L]

(88.6%).22 With this factory-calibrated system, the System previous-generation the product (4.5 minutes, in a study
did not show any marked difference in overall performance without glucose manipulation).2
relative to age, type of diabetes, clinical site, insulin admin- These results have clinical implications for individuals
istration, or hemoglobin HbA1c. The overall MARD of the with diabetes and for the clinicians who treat them. A sensor
System is 9.2% (MARD for pediatric population is 9.7%) with a longer wear period that does not require fingerstick
with the new algorithm, which compares favorably against calibration with its associated burden supports more frequent
the G6 system (MARD of the G6 for the adult and pediatric sensor use with improved glycemic outcomes. Randomized
population are 9.9% and 10.1%, respectively, and the previ- controlled studies and real-world data have revealed better
ous generation of the System had a MARD of 12.0% against glycemic control with the use of the FreeStyle Libre System
YSI reference for the adult population).2,22 over a sustained period of time.9,11,23 This System with the
The new glucose algorithm reduces the time lag for optional alarm feature may benefit groups that have demon-
the System to about two minutes (2.4 minutes for adults strated poor adoption due to alarm fatigue with other CGM
and 2.1 minutes for pediatric population) compared to the systems, wherein some of the alarms are not optional.24
76 Journal of Diabetes Science and Technology 16(1)

Conclusion ORCID iD
The performance of the System was demonstrated by the Shridhara Alva https://orcid.org/0000-0003-3596-2034
overall accuracy of sensor readings and the stability of accu-
rate readings over 14 days of use. The System continues to References
provide the same easy-to-use and comfortable sensor wear 1. Abbott. Abbott Receives CE Mark for FreeStyle® Libre, A
experience for up to 14 days as its previous versions, without Revolutionary Glucose Monitoring System for People with
the need for fingerstick measurements. With the improved Diabetes. Abbott Park, IL: Abbott; 2014.
accuracy and the optional alarm feature, it is anticipated that 2. Bailey T, Bode BW, Christiansen MP, Klaff LJ, Alva
S. The performance and usability of a factory-calibrated
the System will increase the adaption of sensor-based tech-
flash glucose monitoring system. Diabetes Technol Ther.
nology for management of diabetes by people with diabetes. 2015;17(11):787-794.
3. Fokkert MJ, van Dijk PR, Edens MA, et al. Performance of the
Abbreviations FreeStyle Libre Flash glucose monitoring system in patients
BMI, body mass index; CGM, continuous glucose monitoring; CE, with type 1 and 2 diabetes mellitus. BMJ Open Diabetes Res
Conformitè Europëenne; CSII, continuous subcutaneous insulin Care. 2017;5(1):e000320.
infusion; HbA1c, Glycated hemoglobin; MAD, mean absolute dif- 4. Hoss U, Budiman ES, Liu H, Christiansen MP. Feasibility of
ference; MARD, mean absolute relative difference; MDI, multiple factory calibration for subcutaneous glucose sensors in sub-
daily injection; T1D, type 1 diabetes; T2D, type 2 diabetes; SMBG, jects with diabetes. J Diabetes Sci Technol. 2014;8(1):89-94.
self monitoring of blood glucose; YSI, Yellow Spring Instruments. 5. Hoss U, Budiman ES. Factory-calibrated continuous glucose
sensors: the science behind the technology. Diabetes Technol
Acknowledgments Ther. 2017;19(suppl 2):S44-S50.
6. Forlenza GP, Kushner T, Messer LH, Wadwa RP,
The authors acknowledge the study participants and the research
Sankaranarayanan S. Factory-calibrated continuous glucose
staff at the study sites.
monitoring: how and why it works, and the dangers of reuse beyond
approved duration of wear. Diabetes Technol Ther. 2019;21(4):
Declaration of Conflicting Interests 222-229.
The author(s) declared the following potential conflicts of interest 7. Forlenza GP, Argento NB, Laffel LM. Practical considerations
with respect to the research, authorship, and/or publication of this on the use of continuous glucose monitoring in pediatrics and
article: SA, ESB, and HL are employees of Abbott Diabetes Care. older adults and nonadjunctive use. Diabetes Technol Ther.
TB conducts research sponsored by Abbott, Capillary Biomedical, 2017;19(suppl 3):S13-S20.
Dexcom, Diasome, Eli Lilly, Kowa, Lexicon, Medtronic, Medtrum, 8. Kudva YC, Ahmann AJ, Bergenstal RM, et al. Approach to
Novo Nordisk, REMD, Sanofi, Senseonics, Viacyte, vTv Therapeutics, using trend arrows in the FreeStyle Libre flash glucose moni-
Zealand Pharma, and has been a consultant, speaker, and/or advisory toring systems in adults. J Endocr Soc. 2018;2(12):1320-1337.
board member for Abbott, LifeScan, Metronom Health, Novo, Sanofi. 9. Bolinder J, Antuna R, Geelhoed-Duijvestijin P, Kroger J,
RB conducts research sponsored by Abbott, Medtronic, Roche, Weitgasser R. Novel glucose-sensing technology and hypogly-
Novo, Lilly, Sanofi, and Senseonics. caemia in type 1 diabetes: a multicenter, non-masked, random-
KC conducts research sponsored by Medtronic, Dexcom, Abbott, ized control trial. Lancet. 2016;388(10057):2254-2263.
Insulet, and Novo Nordisk, and has been a speaker for Dexcom and 10. Aleppo G, Ruedy KJ, Riddlesworth TD, et al. REPLACE-BG:
Abbott. a randomized trial comparing continuous glucose monitor-
MPC conducts research sponsored by Abbott, Biolinq, Dexcom, Eli ing with and without routine blood glucose monitoring in
Lilly, Novo Nordisk, Medtronic, Insulet, Sanofi Aventis, Merck, adults with well-controlled type 1 diabetes. Diabetes Care.
and Senseonics. 2017;40(4):538-545.
GPF conducts research sponsored by Medtronic, Dexcom, Abbott, 11. Haak T, Hanaire H, Ajjan R, Hermanns N, Riveline JP,
Tandem, Insulet, and Lilly, and has been a consultant, speaker, and/ Rayman G. Flash glucose-sensing technology as a replacement
or advisory board member for Medtronic, Dexcom, Insulet, for blood glucose monitoring for the management of insulin-
Tandem, and Lilly. treated type 2 diabetes: a multicenter, open-label randomized
MK conducts research sponsored by Abbott, Dexcom, Eli Lilly, controlled trial. Diabetes Ther. 2017;8(1):55-73.
Medtronic, NGM Biopharmaceuticals, Tolerion, Pfizer, and 12. Beck RW, Riddlesworth TD, Ruedy K, et al. Continuous glu-
Senseonics. cose monitoring versus usual care in patients with type 2 dia-
DRL conducts research sponsored by Abbott, Dexcom, Drawbridge betes receiving multiple daily insulin injections: a randomized
Health, Eyenuk, Medtronic, Provention Bio, Senseonics, and Trial trial. Ann Intern Med. 2017;167(6):365-374.
Net. 13. Foster NC, Miller KM, Tamborlane WV, Bergenstal RM,
Beck RW. Continuous glucose monitoring in patients with
type 1 diabetes using insulin injections. Diabetes Care. 2016;
Funding 39(6):e81-e82.
The author(s) disclosed receipt of the following financial support 14. Miller KM, Foster NC, Beck RW, et al. Current state of type
for the research, authorship, and/or publication of this article: 1 diabetes treatment in the U.S.: updated data from the T1D
Funding for this study was provided by Abbott Diabetes Care. exchange clinic registry. Diabetes Care. 2015;38(6):971-978.
Alva et al 77

15. Miller KM, Hermann J, Foster N, et al. Longitudinal changes 20. Abbott. New Real-World Data Show that Abbott’s FreeStyle
in continuous glucose monitoring use among individuals with Libre System Delivers Positive Health Outcomes for People
type 1 diabetes: international comparison in the German and with Type 1 and Type 2 Diabetes. Abbott Park, IL: Abbott;
Austrian DPV and U.S. T1D exchange registries. Diabetes 2020.
Care.2020 43(1): e1-e2. 21. Shah VN, Laffel LM, Wadwa RP, Garg SK. Performance of
16. Carlson AL, Mullen DM, Bergenstal RM. Clinical use of a factory-calibrated real-time continuous glucose monitor-
continuous glucose monitoring in adults with type 2 diabetes. ing system utilizing an automated sensor applicator. Diabetes
Diabetes Technol Ther. 2017;19(suppl 2):S4-S11. Technol Ther. 2018;20(6):428-433.
17. Park C, Le QA. The effectiveness of continuous glucose 22. Wadwa RP, Laffel LM, Shah VN, Garg SK. Accuracy of a
monitoring in patients with type 2 diabetes: a systematic factory-calibrated, real-time continuous glucose monitoring
review of literature and meta-analysis. Diabetes Technol Ther. system during 10 days of use in youth and adults with diabetes.
2018;20(9):613-621. Diabetes Technol Ther. 2018;20(6):395-402.
18. Abbott. Abbotts’S FreeStyle® Libre 2, with Optional Real-Time 23. Jangam S, Lang J, Dunn T, Xu Y, Hayter G. 298OR: Sustained
Alarms Secures CE Mark for Use in Europe. Abbott Park, IL: improvement in glycemia following flash glucose monitoring:
Abbott; 2018. an expanded worldwide analysis. Diabetes. 2019;68(suppl 1):
19. Parkes JL, Slatin SL, Pardo S, Ginsberg BH. A new consen- 298OR. https://doi.org/10.2337/db19-298-OR
sus error grid to evaluate the clinical significance of inaccu- 24. Shivers JP, Mackowiak L, Anhalt H. “Turn it off!”: diabetes
racies in the measurement of blood glucose. Diabetes Care. device alarm fatigue considerations for the present and the
2000;23:1143-1148. future. J Diabetes Sci Technol. 2013;7:789-794.