Pediatric vs. Adult Calibration Algorithms

Pediatric vs. Adult Calibration Algorithms

Children and adults wear the same CGM sensors, and the FDA records we read describe no separate calibration algorithm for kids. As of October 2026, the Dexcom G7 and the FreeStyle Libre 3 Plus are both factory-calibrated and cleared for people 2 years and older [1][2]. What changes with age is the evidence behind those clearances: a different reference in the youngest groups, shorter clinic sessions, different wear sites, and fewer sensors lasting to the last day [1][3]. Those details, not a hidden "kids mode," are what you need to read the published pediatric numbers.

One calibration for every cleared age

A factory-calibrated sensor arrives ready to report glucose without fingerstick calibration from the user [2]. FDA's summary for the G7 describes it as factory calibrated, with an optional manual calibration, under one indication for persons 2 years and older [1]. Abbott's Libre 3 Plus sensor goes further: the summary says it "cannot be calibrated by the user" [2]. Neither document mentions an age setting, a pediatric software version or a different glucose conversion for children [1][2].

Here's what surprised us. The one current system we checked that still needs routine fingerstick calibration, the implanted Eversense 365, is labeled only for adults 18 and older [4]. It asks for four calibrations at the start, then one a day through day 13, then one a week [4]. Dexcom's newer G7 15 Day sensor is also limited to adults [5], so among Dexcom's current sensors, the one labeled for children is the 10-day G7 [1].

What the FDA rule asks for in children

The G7 and the Libre Plus sensors are cleared as integrated CGMs (iCGMs), the FDA category for sensors designed to send readings to automated insulin dosing systems and other connected devices [1][2][6]. That rule sets fixed accuracy targets for adults. One example: across the measuring range, the lower 95% confidence bound for readings within 20% of the reference value has to exceed 87% [6].

Children are held to a comparison rather than a fixed bar. The rule asks for data showing "similar accuracy and rate of change performance" in the pediatric population compared with adults, or a justification FDA accepts for why pediatric data aren't needed [6]. It also requires labeling to report sensor performance separately for each intended use population and each insertion site, such as abdomen, arm or buttock [6]. Our iCGM regulatory standards article covers the rest of the rule.

How the pediatric studies were run

A different reference for the youngest children

Accuracy studies pair each sensor reading with a reference glucose value. For adults and older children, both companies measured blood samples on a laboratory analyzer, the YSI 2300 [1][3][7]. For the youngest group, they used fingerstick blood on a home meter instead: ages 2 to 6 in the Dexcom G7 study and ages 2 to 5 in Abbott's study [3][7].

That swap matters when you compare numbers. The iCGM rule describes its reference as an FDA-accepted laboratory method [6]. FDA's guidance for over-the-counter meters recommends that 95% of results fall within 15% of a lab comparator, and 99% within 20% [8]. A meter can meet that guidance and still differ from the lab by several percent, so part of any gap in the youngest group's results may belong to the meter rather than the sensor.

Shorter sessions and smaller samples

FDA's summary applies one pivotal study of the Libre 2 Plus sensor to the Libre 3 Plus as well, because the two sensors are physically identical apart from their transmission dimensions [3]. Adults took part in up to three 10-hour clinic sessions. Teens aged 13 to 17 had two 12-hour sessions, children aged 6 to 12 had one, and children aged 2 to 5 had a single 4-hour session [3]. Study staff deliberately raised and lowered glucose during sessions for participants 11 and older to cover the measuring range; children 10 and younger were only observed [3].

So the youngest children supply far less data. Abbott's 2-to-5 group contributed 135 matched readings from 10 children, against 20,497 readings from 149 adults [3].

What the numbers show by age

MARD, or mean absolute relative difference, is the average percentage gap between sensor readings and the reference; lower means closer agreement. In the G7 studies, arm-worn sensors had a MARD of 8.2% in adults and 8.1% in children and teens aged 7 to 17 [7][9]. Among children aged 2 to 6, measured against fingerstick values, MARD was 9.3% [7]. FDA's summary also reports the share of G7 readings within 20% of the reference: 94.1% for adults and 94.4% for ages 7 to 17 on the arm, 96.4% for ages 2 to 6 on the arm, and 87.4% for ages 2 to 6 on the upper buttocks [1].

For the Libre Plus sensors, 92.5% of adult readings and 92.9% of readings from ages 6 to 17 fell within 20% of the lab value. The 2-to-5 group, compared with a meter, came in at 86.7% [3]. Accuracy also moved across the wear period, and not in the same way for each group. Adults had their highest MARD at the start, 10.0% on days 1 to 3; children aged 6 to 17 had theirs at the end, 10.4% on days 13 to 15 [3]. We compare these figures across brands in Comparative Accuracy and MARD Analysis.

Wear sites and sensor survival

The G7 is cleared for the arm at ages 2 and older, and for the upper buttocks at ages 2 to 6 only [1]. Both G7 studies also tested sensors on the abdomen, with MARDs of 9.1% in adults and 9.0% in ages 7 to 17 [7][9], yet FDA's summary lists only the arm and upper buttocks as G7 wear sites [1]. The Dexcom G6 it was compared against, which Dexcom stopped making after July 1, 2026 and has replaced with its G7 sensors, was labeled for the abdomen at ages 2 and older and the upper buttocks at ages 2 to 17 [1][10]. The Libre Plus sensors have one wear site for every cleared age, the back of the arm [2][3].

Sensor survival is the share of sensors still working on a given day. In the G7 study, 80.5% of adult arm sensors reached day 10, compared with 75.0% for children aged 2 to 17 [1]. For the 16 upper-buttocks sensors worn by children aged 2 to 6, survival by day 10 was 50.0% [1]. Adhesive failures ended 21 of 315 adult sensors and 8 of 145 pediatric sensors early; the rest of the early losses were automatic sensor shutoffs [1]. For the Libre Plus sensors, 83.1% of adult sensors and 76.8% of pediatric sensors lasted to the final day [3]. Which wear site suits a particular child is a decision for the family and the child's care team.

What the public record doesn't show

FDA summaries describe evidence, not code. We found no labeling that explains how either company's software turns a sensor signal into a glucose number, by age or otherwise [1][2], so the idea that pediatric CGMs run separately tuned calibration algorithms remains unsupported. The youngest children's results rest on small groups and a meter-based reference [3][7], and that detail belongs next to any pediatric accuracy figure you read. For how calibration works in general, see Sensor Calibration Algorithms and Data Fusion.