Abstract
Background:
This study evaluates the performance of a 15-day factory-calibrated continuous glucose monitoring sensor used in FreeStyle Libre 2 Plus (Libre 2 Plus) and FreeStyle Libre 3 Plus (Libre 3 Plus) Systems, featuring an improved sensor design to reduce vitamin C interference.
Methods:
Participants aged 2 years and above were enrolled for this study at seven sites in the United States. Depending on their age and bodyweight, participants attended up to three in-clinic sessions where venous blood was obtained for comparator measurement. For 2- to 5-year-olds, only capillary comparator data were collected. Participants aged 11 years and older underwent supervised glycemic manipulation to achieve glucose levels across the sensor’s measurement range. Performance measures included the proportion of continuous glucose monitoring (CGM) values within ±20%/±20 mg/dL of comparator glucose values and mean absolute relative difference (MARD) between CGM and comparator values.
Results:
Of the total 332 participants enrolled in the study, 149 adults and 124 pediatric participants (ages 6-17 years) had paired data for analysis against YSI comparator. Percentages within ±20 mg/dL/20% were 94.2% and 94.0%, and MARDs were 8.2% and 8.1% for the adults and pediatric participants, respectively. For 12 pediatric participants of 2 to 5 years, the percentage within ±20 mg/dL/20% was 86.6%, with an MARD of 11.2% against self-monitoring of blood glucose (SMBG) comparator. The sensor performed well in the hypoglycemic range, with 97.1% and 98.0% of results within ±15 mg/dL of the YSI comparator for the adult and pediatric participants, respectively.
Conclusions:
The Libre 2 Plus and Libre 3 Plus Systems provide accurate glucose results across the dynamic range during the 15-day sensor wear period.
Keywords: FreeStyle Libre 2, continuous glucose monitoring, optional alarm, factory calibration, accuracy
Introduction
The Abbott FreeStyle Libre Flash Glucose Monitoring System (Abbott Diabetes Care, Alameda, CA), introduced in 2014, was the first factory-calibrated continuous glucose monitoring (CGM) device available to people with both type 1 diabetes (T1D) and type 2 diabetes (T2D).1-3 Factory-calibrated CGM devices offer distinct advantages over user-calibrated devices, including reduced burden, reduction in test-strip usage, and removal of user induced calibration inaccuracies.4-6 Improved device accuracy has led to approval for insulin dosing from CGM values (non-adjunctive use), further reducing user burden.7,8 Direct dosing from CGM has been demonstrated to be safe and effective in several randomized clinical trials.9-12 The expansion of CGM utility and reduction in technology burden has contributed to increased adoption of CGM by people with T1D and T2D.13-17
The second generation of the Libre family of products, send data to the reader or smartphone every minute to generate optional hypoglycemic and hyperglycemic alarms. This product secured CE mark approval in 2018 18 and was updated in 2020 with a new algorithm to further improve accuracy. 19 Third generation product, Libre 3 with new features such as smaller sensor, continuous real-time glucose readings automatically delivered to a person’s reader or smartphone every minute and a one-piece applicator was CE marked in 2020. 20 Subsequently, the sensor design was improved to minimize the interference from electroactive compounds by reducing the available electrode area for the electrochemical oxidation of potential interfering compounds. This updated sensor has the same functional features as its predecessor used in Libre 2 and Libre 3 Systems, but the sensor wear duration is extended from 14 to 15 days. The Systems with this updated sensor, the Libre 2 Plus and Libre 3 Plus Systems, obtained FDA clearance in 2023. 21
The study evaluated the accuracy of this modified sensor in participants with T1D or T2D aged 2 years and older. Participants 11 years and older underwent glycemic challenges to obtain sufficient data for the evaluation of sensor performance at low and high glucose concentrations. This study used the Libre 2 Plus device for the accuracy evaluation. The Libre 3 Plus uses the same sensor but with a smaller on-body component. 22 Therefore, the performance of the Libre 2 Plus System is representative of Libre 3 Plus System as well.
Methods
Device Description
The Libre family of products uses a wired enzyme technology with glucose oxidase. The Libre 2 Plus sensor is factory-calibrated and has a wear time of up to 15 days without user calibration. The sensor is designed to send one-minute data to the reader or app while the user can also scan the sensor to get the current glucose, glucose trend and up to eight hours of historic glucose values. The Libre 3 Plus sensor sends glucose data every minute to the reader or app for both optional alarms, and for providing current glucose, glucose trends and historic glucose values to the user. Both Systems also have optional hypoglycemic and hyperglycemic threshold alarms.
Study Design
The sensor was evaluated in a prospective multicenter study enrolling 332 participants aged 2 years and older with T1D or T2D at seven clinical sites in the United States. All participants were required to wear the sensor for up to 15 days and perform four blood glucose (BG) measurements daily. All BG measurements were performed using the reader’s built-in blood glucose meter, that uses Precision Neo test strips. Participants received no training on the devices (but were trained on study procedures) and inserted one sensor on the back of each of their arms following the instructions for use. For pediatric participants, either the parent or the participant inserted the sensor. Only sensors that failed within the first hour after insertion were replaced.
The participants aged 11 and above were required to have intensive insulin therapy (either multiple daily injections or continuous subcutaneous insulin infusion) with known insulin sensitivity factor and underwent controlled manipulation of glucose levels during at least one of the in-clinic sessions to generate more data at high and low glucose levels. The manipulation included controlling food intake and insulin administration to achieve and maintain glucose levels of <70 or >300 mg/dL for approximately one hour. Glucose levels permitting, hypoglycemic induction was performed prior to the hyperglycemic induction in the same session. Exclusion criteria included pregnancy, anemia, or any condition, per investigator discretion, that could place the participant at risk by glucose manipulation. No exclusions were made based on any other concomitant medications or supplements.
Adult participants were scheduled for up to three in-clinic sessions of 10 hours and pediatric participants aged 6 years and older were scheduled for up to two in-clinic sessions of up to 10 hours. The in-clinic sessions covered 12 of the 15 sensor wear days, including days 1 to 3 (beginning), 5 to 7 (early middle), 9 to 11 (late middle), and 13 to 15 (end). All sessions included comparison of sensor readings to plasma venous glucose concentrations using a laboratory comparator method (Yellow Springs Instrument YSI 2300 (YSI); YSI, Inc., Yellow Springs, OH). Participants had their venous blood drawn every 15 minutes for the duration of each clinic session for YSI measurement by a study staff. Sampling frequency was increased to every five minutes for up to an hour when the glucose concentration was <70 or >250 mg/dL. No more than approximately 64 samples were collected in one in-clinic session. Participants or caregivers/study staff scanned both sensors with the paired reader immediately following comparator sample collection. Each sample was centrifuged within 15 minutes of the blood draw and tested on YSI within 15 minutes thereafter. Each blood draw sample was assayed on the YSI in duplicate and the average of the readings used. A heating pad was applied to the arm in which the IV was placed. Pediatric participants aged 2 to 5 years underwent one in-clinic session of up to four hours where capillary BG measurements were performed approximately every 15 minutes. The CGM devices were masked to participants and clinic staff for the entire sensor wear. Data from the sensors were uploaded using a proprietary software, at the end of the sensor wear period. Kaplan-Meier estimation method was used for the sensor survival analysis. 23
The lag between the sensor glucose reading and the YSI reference was evaluated by performing least square linear regression of the difference between the sensor and YSI readings versus the sensor rate of change. The sensor rate of change was calculated as the instantaneous rate of change (ie, using the sensor reading matched to YSI, along with the previous and following values, covering a total of 30 minutes). The slope of the regression line is the mean lag time.
Questionnaires with 5-point Likert scale were administered during the sensor application and removal visits to collect study participants’ assessment of sensor’s ease of use and wear pain. The study was registered with clinicaltrials.gov (NCT# 05251116).
Data Analysis
The first inserted sensor with YSI-sensor pair was used for the analysis of accuracy performance while the second applied sensor was used for the assessment of precision. One minute sensor values were paired with YSI values by choosing the sensor value that was closest in time to the YSI blood draw, but no more than five minutes before or after the YSI blood draw. Only matched pairs where the sensor results were within the reportable range (40-500 mg/dL) with pairable comparator results were used for the performance evaluation included in the analyses. The agreement levels were calculated relative to a glucose concentration difference (in mg/dL) when the comparator glucose value was up to 80 mg/dL and relative to a normalized concentration difference (in %) otherwise, and were evaluated at three different ranges: ±15%/±15 mg/dL, ±20%/±20 mg/dL, and ±40%/±40 mg/dL. The mean absolute relative difference (MARD) was calculated as the absolute value of the average percent difference between the paired sensor and comparator glucose values. Consensus Error Grid and DTS Error Grid analysis was performed for paired glucose data.24,25
The total drift of the sensitivity over the wear period was assessed by performing linear regression on the paired readings of relative difference between the sensor and SMBG against sensor elapsed time. The between sensor precision was calculated as the coefficient of variation (CV) from the paired historic glucose readings from two sensors worn simultaneously with matched wear time. Overall CV was calculated as arithmetic mean over the pairs. The ability of the Systems to alarm appropriately when alerts were set at different thresholds was assessed by comparing sensor results to YSI measurements within a 15-minute time window at different low and high glucose thresholds, and determining whether an alert would have been generated. At each threshold, the true alarm rate (whether YSI agreed with the device when the device alerted) and detection rate (whether the device alerted when YSI was within the threshold) were calculated. The false alarm rate and missed detection rates were calculated as (100%-true alarm rate) and (100%-detection rate), respectively.
All analyses were performed using SAS 9.4 Software.
Results
Of the 332 participants enrolled in the study, thirty-three (n = 33) participants were screen failed (29 of 33 due to hemoglobin levels below the normal range for their age and gender, one (1) due to previous DKA, three (3) due to other conditions that led the investigator to decide that it was not safe for the participant to participate in the study), and six participants (n = 6) withdrew from the study prior to sensor application. One participant was lost to follow up and did not return the sensors, thus had no CGM paired with comparator results. Of the 19 participants who had CGM paired with BG but not YSI comparator, 12 participants were between ages 2 and 5 years, while six participants between ages 6 and 17 years either had both sensors fall off prior to the in-clinic session or missed the in-clinic session due to illness, and one adult participant withdrew from the study prior to the in-clinic session.
Of the 273 participants who had paired YSI comparator results, 124 participants were between ages 6 and 17 years, and 149 participants were 18 years and older. Twelve pediatric participants aged 2 to 5 years were evaluable for accuracy assessment against BG comparator. All participants who had a sensor applied were included in the safety assessment. Demographics and baseline characteristics of the evaluable participants in the study are provided in Table 1. The participant population had a good spread of HbA1c values covering a wide range and represented both CSII and MDI users. Majority of the study participants (64.7% of adult and 76.8% of pediatric) were using the pumps for managing their diabetes. The investigational sensors used for the accuracy evaluation were not integrated into the pump. The BMI ranged from 19.4 to 54.3 kg/m2 for adults and 14.3 to 43.1 kg/m2 for the pediatric participants. Greater than 85% of the adult participants had T1D, while all pediatric participants had T1D.
Table 1.
(a) Demographics and (b) Characteristics of Study Participants in Adult and Pediatric Populations.
| (a). | |||||
|---|---|---|---|---|---|
| Demographic | Adult Population | Pediatric Population | |||
| N | % | N | % | ||
| Sex | Female | 80 | 53.3 | 58 | 40.8 |
| Male | 70 | 46.7 | 84 | 59.2 | |
| Race | White—Not Hispanic or Latino | 116 | 77.3 | 102 | 71.8 |
| White—Hispanic or Latino | 24 | 16.0 | 29 | 20.4 | |
| Asian | 3 | 2.0 | 1 | 0.7 | |
| Other | 3 | 2.0 | 1 | 0.7 | |
| Black or African American | 2 | 1.3 | 6 | 4.2 | |
| Native Hawaiian or Pacific Islander | 0 | 0 | 1 | 0.7 | |
| Type of Diabetes | Type 1 | 128 | 85.3 | 142 | 100 |
| Type 2 | 22 | 14.7 | 0 | 0 | |
| Insulin Pump Use | Yes | 97 | 64.7 | 109 | 76.8 |
| No | 53 | 35.3 | 33 | 23.2 | |
| HbA1c | HbA1c < 7% | 57 | 38.0 | 20 | 14.3 |
| 7% ≤ HbA1c ≤ 8.5% | 64 | 42.7 | 67 | 47.9 | |
| HbA1c > 8.5% | 29 | 19.3 | 53 | 37.9 | |
| (b). | |||||
|---|---|---|---|---|---|
| Characteristic | Adult Population | Pediatric Population | |||
| Mean ± SD | Range | Mean ± SD | Range | ||
| Age (years) | 43.3 (16.8) | 18 to 76 | 12.7 (3.7) | 2 to 17 | |
| Weight | Pounds | 192.1 (48.7) | 113.0 to 388.6 | 126.1 (49.5) | 23.7 to 273.4 |
| Kilograms | 87.1 (22.1) | 51.3 to 176.3 | 57.2 (22.4) | 10.8 to 124.0 | |
| Height | Inches | 67.6 (3.9) | 58 to 76 | 61.9 (8.0) | 31 to 75 |
| Meters | 1.72 (0.10) | 1.47 to 1.93 | 1.57 (0.20) | 0.79 to 1.91 | |
| Body Mass Index (BMI) | 29.5 (6.6) | 19.4 to 54.3 | 22.2 (5.5) | 14.3 to 43.1 | |
| Duration of diabetes (years) | 21.9 (13.2) | 1.0 to 53.1 | 5.5 (3.2) | 0.2 to 15.8 | |
| Duration of insulin use (years) | 19.8 (13.6) | 1.0 to 53.0 | 5.4 (3.2) | 0.2 to 16.0 | |
| Total number of injections per day (No. of participants = 38) |
4.6 (1.1) | 3 to 9 | 4.6 (0.8) | 4 to 6 | |
| HbA1c (%) | 7.5 (1.6) | 5.0 to 14.7 | 8.4 (1.6) | 6.0 to 15.7 | |
The majority of the study participants reported that the sensor application was easy (98.7% adults and 97.9% pediatrics) and painless (94.0% adults and 79.6% pediatrics). At the end of the sensor wear, most participants reported that wearing the sensor was painless (97.4% adults and 90.8% pediatrics).
The overall performance by the age groups is presented in Table 2. There were a total of 20 619 matched data pairs in the adult group and 7075 matched data pairs in the pediatric group, against YSI, and 477 matched data pairs against BG in the 2- to 5-year-old participants. The MARD of the adult group was 8.2%, and for the 6- to 17-year-old pediatric group was 8.1%. For the participants aged 2 to 5 years, the MARD against BG comparator was 11.2%. The analyses showed that 94.2% and 94.0% of sensor readings were within 20% or 20 mg/dL of YSI results for the adult and pediatric groups, respectively, while 86.6% of the results were within 20% or 20 mg/dL of the BG comparator for the 2- to 5-year-old group. The subgroup analysis did not show any marked difference in overall performance relative to age (P value = .6445), type of diabetes (P value = .2689), clinical site (P value = .3473), insulin administration (P value = .4216), or HbA1c (P value = .3554). Consensus Error Grid analysis 19 showed 94.3% and 93.0% of the data in the A zone and 99.9% and 100.0% of the data in the A+B zone of the error grid for the adult and pediatric groups, respectively. DTS Error Grid analysis showed 92.1% and 92.2% of the data in the No Risk zone of the error grid for the adult and pediatric groups, respectively, and 99.9% of the data in the No Risk or Mild Risk zone for both population groups.
Table 2.
Overall Performance by Participant Age Group Against YSI Comparator.
| Participant group | Participants | Matched pairs (n) | % within ±15%/±15 mg/dL | % within ±20%/±20 mg/dL | % within ±40%/±40 mg/dL | MARD, % | MRD,% |
|---|---|---|---|---|---|---|---|
| Adults (ages 18+) | 149 | 20 619 | 89.3 | 94.2 | 99.4 | 8.2 | -2.8 |
| Pediatric (ages 6-17) | 124 | 7075 | 88.4 | 94.0 | 99.5 | 8.1 | -4.5 |
| Pediatric* (ages 2-5) | 12 | 477 | 78.0 | 86.6 | 97.3 | 11.2 | -0.8 |
Results based on comparing with SMBG comparator.
Relative agreement at different YSI glucose ranges is presented in Table 3. Both the adult and pediatric studies exhibited similar agreement rates across glucose value ranges. The accuracy at lower glucose levels was >97% within 20 mg/dL for both adults (Table 3a) and pediatric (Table 3b) participants. Performance <70 mg/dL is also provided for comparison in Table 3c.
Table 3.
Accuracy Performance at Different Glucose Levels and Rate of Changes. (a) Data From the Adult Population, (b) Data From the Pediatric Population, (c) Data From the Adult and the Pediatric Populations in the Medically Relevant Glucose Ranges, and (d) Data From the Adult and the Pediatric Populations in Different Glucose Rate of Change Ranges.
| (a). | |||||
|---|---|---|---|---|---|
| YSIref Glucose level, mg/dL [mmol/L] ǂ | % Within ±15%/±15 mg/dL | % Within ±20%/±20 mg/dL | % Within ±40%/±40 mg/dL | MAD, mg/dL/MARD, %* | N |
| 40–50 [2.2–2.8] |
92.2 | 97.7 | 100.0 | 8.5 | 257 |
| 51–80 [2.8–4.4] |
94.1 | 97.6 | 99.4 | 6.1 | 4319 |
| 81–180 [4.5–10.0] |
81.6 | 89.6 | 98.7 | 9.6 | 7070 |
| 181–300 [10.0–16.7] | 91.8 | 95.4 | 99.8 | 6.7 | 5793 |
| 301–400 [16.7–22.2] | 95.0 | 97.7 | 99.6 | 5.5 | 3093 |
| 401–500 [22.3–27.8] | 90.8 | 92.0 | 100.0 | 6.2 | 87 |
| Overall | 89.3 | 94.2 | 98.3 | 8.2 | 20 619 |
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. ǂAccuracy results for glucose values ≤80 mg/dL are in mg/dL. Overall MARD reported is calculated as mean absolute relative difference across 40 to 500 mg/dL.
| (b). | |||||
|---|---|---|---|---|---|
| YSIref Glucose Level, mg/dL [mmol/L] ǂ | % Within ±15%/±15 mg/dL | % Within ±20%/±20 mg/dL | % Within ±40%/±40 mg/dL | MAD, mg/dL/ MARD, %* | N |
| 40–50 [2.2–2.8] |
100.0 | 100.0 | 100.0 | 6.2 | 40 |
| 51–80 [2.8–4.4] |
93.4 | 97.2 | 99.7 | 5.9 | 1045 |
| 81–180 [4.5–10.0] |
79.4 | 88.2 | 96.2 | 10.1 | 2786 |
| 181–300 [10.0–16.7] | 93.1 | 97.6 | 99.5 | 6.6 | 2786 |
| 301–400 [16.7–22.2] | 96.8 | 98.5 | 98.9 | 5.3 | 975 |
| 401–500 [22.3–27.8] | 100.0 | 100.0 | 100.0 | 4.1 | 51 |
| Overall | 88.4 | 94.0 | 98.2 | 8.1 | 7075 |
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. ǂAccuracy results for glucose values ≤80 mg/dL are in mg/dL. Overall MARD reported is calculated as mean absolute relative difference across 40 to 500 mg/dL.
| (c). | ||||||||
|---|---|---|---|---|---|---|---|---|
| YSIref Glucose Level, (mg/dL) [mmol/L] ǂ |
Adult Population | Pediatric Population | ||||||
| No. Pair | % Within ±15%/±15 mg/dL | % Within ±20%/±20 mg/dL | % Within ±40%/±40 mg/dL | No. Pair | % Within ±15%/±15 mg/dL | % Within ±20%/±20 mg/dL | % Within ±40%/±40 mg/dL | |
| <70 [<3.9] |
3259 | 97.1 | 98.9 | 99.7 | 693 | 98.0 | 99.9 | 100.0 |
| 70-180 [3.9-10.0] |
8387 | 82.3 | 90.4 | 98.9 | 3178 | 80.2 | 88.7 | 99.1 |
| >180 [>10.0] |
8973 | 92.9 | 96.2 | 99.9 | 3204 | 94.4 | 97.9 | 99.9 |
Accuracy results for glucose values <70 mg/dL are in mg/dL.
| (d). | ||||||||
|---|---|---|---|---|---|---|---|---|
| YSIref Glucose ROC, (mg/dL/min) | Adult Population | Pediatric Population | ||||||
| No. Pair | % Within ±15%/±15 mg/dL | % Within ±20%/±20 mg/dL | % Within ±40%/±40 mg/dL | No. Pair | % Within ±15%/±15 mg/dL | % Within ±20%/±20 mg/dL | % Within ±40%/±40 mg/dL | |
| <-2 | 424 | 81.6 | 89.6 | 99.1 | 121 | 92.6 | 95.9 | 100.0 |
| -2 to -1 | 1586 | 88.3 | 93.7 | 99.7 | 600 | 89.2 | 95.3 | 99.7 |
| -1 to 1 | 14 412 | 90.6 | 95.0 | 99.5 | 4888 | 89.0 | 94.4 | 99.6 |
| 1 to 2 | 2234 | 88.8 | 94.4 | 99.6 | 738 | 88.6 | 93.4 | 99.2 |
| >2 | 1372 | 79.7 | 88.1 | 98.7 | 468 | 79.3 | 88.2 | 98.7 |
| NA* | 591 | 89.0 | 92.9 | 98.8 | 260 | 87.7 | 93.5 | 100.0 |
Rate of change not available.
Percentage within 20% or 20 mg/dL and MARD performance throughout the wear duration was assessed for both adult and pediatric participants and is presented in Table 4. Percentage of the data within 20% or 20 mg/dL for the adult and pediatric participants on first day of the sensor wear was 83.5% and 90.4%, respectively. MARD on the first day of the sensor wear for the adult and pediatric data were 12.8% and 9.4% respectively. The sensor performance remained consistent across the wear period.
Table 4.
Percent Within 20%/20 mg/dL and Mean Absolute Relative Difference Performance at Different Wear Periods for Adult and Pediatric Populations.
| Wear Period (days) | Within ±20%/±20 mg/dL | Adult Population |
Within ±20%/±20 mg/dL | Pediatric Population |
||||
|---|---|---|---|---|---|---|---|---|
| Outside ±40%/±40 mg/dL | Mean Absolute Relative Difference, MARD, % | N | Outside ±40%/±40 mg/dL | Mean Absolute Relative Difference, MARD, % | N | |||
| Beginning (1-3) (Day 1)* | 90.4 (83.5) |
0.9 (1.9) |
10.0 (12.8) |
5460 (2035) |
91.7 (90.4) |
0.5 (0.6) |
9.0 (9.4) |
2660 (1078) |
| Early Middle (5-7) | 96.5 | 0.2 | 7.2 | 5057 | 97.5 | 0.1 | 6.8 | 2282 |
| Late Middle (9-11) | 95.5 | 0.7 | 7.7 | 5163 | 97.1 | 0.2 | 6.9 | 1221 |
| End (13-15) | 94.8 | 0.4 | 7.8 | 4939 | 87.5 | 2.0 | 10.4 | 912 |
| Overall | 94.2 | 0.6 | 8.2 | 20 619 | 94.0 | 0.5 | 8.1 | 7075 |
Performance on day 1 is provided in parenthesis with the data for the beginning phase of the sensor wear.
The total drift in the sensor signal estimated over the 15-day wear duration was 4.2% and 2.6% for the adult and pediatric populations, respectively. Estimated probability of sensor surviving 15 days was 83.1% and 76.8% for adults and pediatric participants, respectively. Of the sensors that did not survive, 2.7% for adults and 2.1% for pediatric population were due to the sensors terminating early, and the remaining were due to sensors getting dislodged during wear.
The alarm performance when alarms were set at different thresholds at low (60, 70, 80, and 90 mg/dL) and high (120, 140, 180, 200, 220, 240, and 300 mg/dL) glucose threshold levels are presented in Table 5. Every CGM or YSI value beyond these glucose threshold levels was assessed separately.
Table 5.
Glycemic Alarms From the Adult and Pediatric Populations.
| Threshold, mg/dL [mmol/L] | Adult Population |
Pediatric Population (6-17 years) |
|||
|---|---|---|---|---|---|
| Detection Rate,%/N | True Alarm Rate,%/N | Detection Rate, %/N | True Alarm Rate, %/N | ||
| Hypoglycemic range | 60 mg/dL [3.3 mmol/L] |
84.5/1376 | 71.1/9756 | 87.6/275 | 58.8/2762 |
| 70 mg/dL [3.9 mmol/L] |
95.5/3451 | 84.6/23 078 | 98.6/735 | 74.3/6129 | |
| 80 mg/dL [4.4 mmol/L] |
98.0/4655 | 90.8/33 676 | 98.6/1104 | 82.8/9659 | |
| 90 mg/dL [5.0 mmol/L] |
98.8/5525 | 92.2/42 322 | 99.7/1434 | 88.3/13111 | |
| Hyperglycemic range | 120 mg/dL [6.7 mmol/L] |
97.8/13 213 | 99.3/96 119 | 97.2/4848 | 99.4/34770 |
| 140 mg/dL [7.8 mmol/L] |
98.0/11 729 | 99.2/83 016 | 97.2/4271 | 99.2/29884 | |
| 180 mg/dL [10.0 mmol/L] |
98.0/9337 | 98.8/61 513 | 9739/3355 | 99.1/21901 | |
| 200 mg/dL [11.1 mmol/L] |
98.0/8388 | 98.5/53 287 | 98.0/3033 | 99.2/18875 | |
| 220 mg/dL [12.2 mmol/L] |
97.8/7615 | 98.4/45 745 | 96.9/2756 | 98.7/15959 | |
| 240 mg/dL [13.3 mmol/L] |
97.2/6902 | 98.9/38 393 | 96.0/2452 | 98.4/12841 | |
| 300 mg/dL [16.7 mmol/L] |
91.2/3369 | 94.8/16 594 | 92.2/1100 | 97.5/5167 | |
The time lag between the venous comparator and the sensor results was calculated for the adult and pediatric populations as 1.7 ± 3.5 and 0.9 ± 3.9 minutes, respectively. Mean CV between the sensors in the adult and pediatric populations were 5.7% and 6.1%, respectively.
Discussion
Libre family of products have allowed easy access to the sensing technology with over six million patients using the device to manage their diabetes. 26 The new design has reduced interference from electrochemically active compounds compared with currently marketed Libre family of products. 21 The form factors of Libre 2 Plus and Libre 3 Plus are identical to their previous generations, maintaining the ease of use and comfort to wear. The sensors are factory-calibrated without the need for fingerstick calibration by the user.
The new sensor has demonstrated a performance of 94.2% and 94.0% of the results within ±20% or ±20 mg/dL of YSI comparator for the adult and pediatric populations, respectively, with over 27 500 paired datapoints. The sensor performed well in the hypoglycemic range with over 97.1% of the results for the adult population and 98.0% of the results for pediatric population being within 15 mg/dL of the YSI comparator. Accuracy of the sensor is stable over the sensor wear period. There was no significant difference in overall performance relative to age, type of diabetes, clinical site, insulin administration or HbA1c. The overall MARD is 8.2% (MARD for pediatric population is 8.1%) with the new design.
The lag time of the sensor is less than two minutes (1.7 ± 3.5 minutes for adults and 0.9 ± 3.9 minutes for pediatrics), which is comparable to the previous sensor design (2.4 ± 4.6 minutes for adults and 2.1 ± 5.0 minutes for pediatrics). 19 The sensor is stable across the wear time and demonstrated accuracy across the measurement range, including at low and high glucose levels. These results have clinical implications for individuals with diabetes and their clinicians, making treatment decisions and /or pairing with automated insulin delivery systems. A sensor with a longer wear period that does not require fingerstick calibration supports more frequent and easier access to glucose data with improved glycemic outcomes. Randomized controlled studies and real-world data have revealed better glycemic control with the use of the Libre System over a sustained period of time.9,11,27
Conclusion
The performance of the new sensor in the Libre 2 Plus and Libre 3 Plus Systems was demonstrated by the overall accuracy of sensor readings and the stability of accurate readings over 15 days of use. The sensor continues to provide the same easy-to-use and comfortable sensor wear experience as the previous versions for up to 15 days, without the need for fingerstick calibration or measurements. With reduced interference, it is anticipated that the new sensor will increase the adoption of sensor-based technology for management of diabetes by people with diabetes, including integration with automated insulin delivery systems.
Acknowledgments
The authors acknowledge the study participants and the research staff at the study sites.
Footnotes
Abbreviations: BMI, body mass index; CGM, continuous glucose monitoring; CE, Conformitè Europëenne; CSII, continuous subcutaneous insulin infusion; HbA1c, Glycated hemoglobin; MAD, mean absolute difference; MARD, mean absolute relative difference; MDI, multiple daily injection; T1D, type 1 diabetes; T2D, type 2 diabetes; SMBG, self-monitoring of blood glucose; YSI, Yellow Spring Instruments.
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: S.A., MN and HL are employees of Abbott Diabetes Care. A.B. is an employee of Iowa Diabetes. BB is an employee of Atlanta Diabetes Associates. RB is a prior employee of Rainier Clinical Research Center. KC is an employee of Sansum Diabetes Research Institute. MK is an employee of Diabetes & Glandular Disease Clinic. DRL is an employee of Rocky Mountain Diabetes Center. HT in an employee of Texas Diabetes and Endocrinology.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this study was provided by Abbott Diabetes Care.
ORCID iDs: Shridhara Alva 
https://orcid.org/0000-0003-3596-2034
Kristin Castorino
https://orcid.org/0000-0001-6386-9203
References
- 1. Abbott receives CE mark for FreeStyle® Libre, A revolutionary glucose monitoring system for people with diabetes. ABBOTT PARK, Ill. Published 2014. Accessed September 3, 2014. https://abbott.mediaroom.com/2014-09-03-Abbott-Receives-CE-Mark-for-FreeStyle-Libre-a-Revolutionary-Glucose-Monitoring-System-for-People-with-Diabetes
- 2. Bailey T, Bode BW, Christiansen MP, Klaff LJ, Alva S. The performance and usability of a factory-calibrated flash glucose monitoring system. Diabetes Technol Ther. 2015;17(11):787-794. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Fokkert MJ, van Dijk PR, Edens MA, et al. Performance of the FreeStyle Libre flash glucose monitoring system in patients with type 1 and 2 diabetes mellitus. BMJ Open Diabetes Res Care. 2017;5(1):e000320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Hoss U, Budiman ES, Liu H, Christiansen MP. Feasibility of factory calibration for subcutaneous glucose sensors in participants with diabetes. J Diabetes Sci Technol. 2014;8(1):89-94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Hoss U, Budiman ES. Factory-calibrated continuous glucose sensors: the science behind the technology. Diabetes Technol Ther. 2017;19(S2):S44-S50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Forlenza GP, Kushner T, Messer LH, Wadwa RP, Sankaranarayanan S. Factory-calibrated continuous glucose monitoring: how and why it works, and the dangers of reuse beyond approved duration of wear. Diabetes Technol Ther. 2019;21(4):222-229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Forlenza GP, Argento NB, Laffel LM. Practical considerations on the use of continuous glucose monitoring in pediatrics and older adults and nonadjunctive use. Diabetes Technol Ther. 2017;19(S3):S13-S20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Kudva YC, Ahmann AJ, Bergenstal RM, et al. Approach to using trend arrows in the FreeStyle Libre flash glucose monitoring systems in adults. J Endocr Soc. 2018;2(12):1320-1337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Bolinder J, Antuna R, Geelhoed-Duijvestijin P, Kroger J, Weitgasser R. Novel glucose-sensing technology and hypoglycaemia in type 1 diabetes: a multicenter, non-masked, randomized control trial. The Lancet. 2016;10057:2254-2263. [DOI] [PubMed] [Google Scholar]
- 10. Aleppo G, Ruedy KJ, Riddlesworth TD, et al. REPLACE-BG: a randomized trial comparing continuous glucose monitoring with and without routine blood glucose monitoring in adults with well-controlled type 1 diabetes. Diabetes Care. 2017;40(4):538-545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Haak T, Hanaire H, Ajjan R, Hermanns N, Riveline JP, Rayman G. Flash glucose-sensing technology as a replacement for blood glucose monitoring for the management of insulin-treated type 2 diabetes: a multicenter, open-label randomized controlled trial. Diabetes Ther. 2017;8(1):55-73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Beck RW, Riddlesworth TD, Ruedy K, et al. Continuous glucose monitoring versus usual care in patients with type 2 diabetes receiving multiple daily insulin injections: a randomized trial. Ann Intern Med. 2017;167(6):365-374. [DOI] [PubMed] [Google Scholar]
- 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;39(6):e81-e82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Miller KM, Foster NC, Beck RW, et al. Current state of type 1 diabetes treatment in the U.S.: updated data from the T1D exchange clinic registry. Diabetes Care. 2015;38(6):971-978. [DOI] [PubMed] [Google Scholar]
- 15. Miller KM, Hermann J, Foster N, et al. Longitudinal changes in continuous glucose monitoring use among individuals with type 1 diabetes: international comparison in the German and Austrian DPV and U.S. T1D exchange registries. Diabetes Care. 2019;43(1):e1-e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Carlson AL, Mullen DM, Bergenstal RM. Clinical use of continuous glucose monitoring in adults with type 2 diabetes. Diabetes Technol Ther. 2017;19(S2):S4-S11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Park C, Le QA. The effectiveness of continuous glucose monitoring in patients with type 2 diabetes: a systematic review of literature and meta-analysis. Diabetes Technol Ther. 2018;20(9):613-621. [DOI] [PubMed] [Google Scholar]
- 18. Abbotts’s FreeStyle® Libre 2, with optional real-time alarms secures CE mark for use in Europe. Abbott Park, IL. Published 2018. Accessed October 1, 2018. https://abbott.mediaroom.com/2018-10-01-Abbott-s-FreeStyle-R-Libre-2-with-Optional-Real-Time-Alarms-Secures-CE-Mark-for-Use-in-Europe
- 19. Alva S, Bailey T, Brazg R, et al. Accuracy of a 14-day factory-calibrated continuous glucose monitoring system with advanced algorithm in pediatric and adult population with diabetes. J Diabetes Sci Technol. 2022;16(1):70-77. doi: 10.1177/1932296820958754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. FreeStyle Libre 3 world’s smallest sensor is here. Abbott Park, IL. Published 2020. https://www.abbott.com/corpnewsroom/strategy-and-strength/freeStyle-libre-3-worlds-smallest-sensor-is-here.html
- 21. 510(k) premarket notification. Accessed August 1, 2024. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K222447
- 22. Alva S, Brazg R, Castorino K, Kipnes M, Liljenquist DR, Liu H. Accuracy of the third generation of a 14-day continuous glucose monitoring system. Diabetes Ther. 2023;14(4):767-776. doi: 10.1007/s13300-023-01385-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Kaplan EL, Meier P. “Nonparametric estimation from incomplete observations”. J Amer Statist Assoc. 1958;53(282):457-481. doi: 10.2307/2281868. [DOI] [Google Scholar]
- 24. Parkes JL, Slatin SL, Pardo S, Ginsberg BH. A new consensus error grid to evaluate the clinical significance of inaccuracies in the measurement of blood glucose. Diabetes Care. 2000;23(8):1143-1148. [DOI] [PubMed] [Google Scholar]
- 25. Klonoff DC, Freckmann G, Pleus S, et al. The diabetes technology society error grid and trend accuracy matrix for glucose monitors. J Diabetes Sci Technol. 2024;18(6):1346-1361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Abbott Enters US. Consumer biowearables market with lingo and Libre Rio. Abbott Park, IL. Published 2024. https://www.abbott.com/corpnewsroom/strategy-and-strength/abbott-enters-us-consumer-biowearables-market-with-lingo-and-libre-rio.html
- 27. Jangam S, Lang J, Dunn T, Xu Y, Hayter G. Sustained improvement in glycemia following flash glucose monitoring: an expanded worldwide analysis. Diabetes. 2019;68(S1):298OR. [Google Scholar]