iCGM Regulatory Standards

iCGM Regulatory Standards

An integrated continuous glucose monitor, or iCGM, is a sensor in an FDA device category with its own rules: Class II, regulation 21 CFR 862.1355, product code QBJ. The category began on March 27, 2018, when the FDA granted Dexcom's De Novo request for the Dexcom G6 [1]. A sensor that meets its special controls can be labeled to replace fingerstick testing for treatment decisions and to send readings to other devices, including automated insulin delivery (AID) systems [2]. Two parts of the rule matter to anyone who tests with a meter. It judges accuracy against the same kind of lab reference that strip makers use, and the iCGM labeling we read still sends users back to a meter in named situations.

What the category requires

The FDA defines an iCGM as a system that measures glucose "continuously or frequently" and is designed to "reliably and securely transmit glucose measurement data to digitally connected devices" [2]. The special controls work like a checklist. A maker needs clinical accuracy data from people who represent the intended users, with performance that holds across the whole wear period. It also has to test interfering substances such as medications, run a usability study, block a sensor from working past its claimed wear time, and deliver readings "at clinically meaningful time intervals" [2]. The labeling must then report accuracy in five glucose bands, along with trend accuracy, data gaps and alert rates [2]. Our article on AID algorithms and CGM requirements walks through those accuracy targets one range at a time.

The sensor that started the category is now a legacy product. Dexcom says it will no longer manufacture the G6 after July 1, 2026, and points G6 users to the Dexcom G7 or G7 15 Day [3].

Which sensors are iCGMs

As of October 2026, the FDA's 510(k) records list 18 decisions under product code QBJ. Every one belongs to Dexcom or Abbott, from the 2018 G6 De Novo to a June 2026 clearance covering the Dexcom G7 and G7 15 Day [4]. The first G7 clearance, in December 2022, sits under the same regulation [4] [5], as does Abbott's April 2024 clearance for the FreeStyle Libre 2 Plus and Libre 3 Plus sensors [6].

Here's what surprised us: being an iCGM and being cleared for AID aren't the same thing. That Abbott clearance also covered the older Libre 2 and Libre 3 sensors, which were indicated for connected devices only "where the user manually controls actions for therapy decisions" and "must not be used with automated insulin dosing (AID) systems." The Plus sensors, under the same clearance, are indicated for AID [6]. Abbott kept the older sensors available only through September 30, 2025 [7].

Implanted sensors took their own route. Senseonics' De Novo request for the Eversense AP CGM System created a sister category, 21 CFR 862.1357, for an iCGM "with sensor containing dexamethasone," a steroid meant to reduce inflammation at the insertion site [8].

How the accuracy bar compares with strip guidance

FDA's guidance for over-the-counter meters and strips, issued September 29, 2020, asks for 95% of results within ±15% of the reference and 99% within ±20% across the claimed measuring range. The study should include at least 350 people, each testing themselves with only the draft instructions [9]. The companion guidance for prescription point-of-care meters, the ones clinicians use, sets a tighter bar: 95% within ±12% (±12 mg/dL below 75 mg/dL) and 98% within ±15% (±15 mg/dL below 75) [10].

The iCGM rule asks a different question. Across its range, more than 87% of adult readings must land within ±20% of the lab value, judged by the lower one-sided 95% confidence bound rather than the raw percentage [2]. It also allows zero readings that are wrong in the most dangerous way: a sensor value below 70 mg/dL can't pair with a lab value above 180, or the reverse [2]. Those targets look looser than the strip numbers, but the jobs differ. A strip reads one drop of capillary blood. A CGM reads interstitial fluid, the fluid around cells under the skin, every few minutes for days [2].

The legal weight differs as well. Both strip guidances are labeled "Contains Nonbinding Recommendations," and both say a meter that misses the targets may still be found substantially equivalent if other features make up for it [9] [10]. The iCGM text says a system "must comply" with its special controls [2].

The overlap we didn't expect sits in the definitions. Both documents describe the reference in nearly the same words: a laboratory-based glucose method that is precise and accurate and "traceable to a higher order," such as an internationally recognized reference material or method [2] [9]. For more on how strips reach the market, see our article on 510(k) pathways for [test strips](/diabetic-test-strips/regulatory-pathways-510k-for-test-strips/).

What "non-adjunctive" means for fingerstick checks

An adjunctive CGM supplements a meter. A non-adjunctive one is labeled to stand in for it. On December 20, 2016, the FDA approved a supplement saying the Dexcom G5 Mobile was "designed to replace fingerstick blood glucose testing for diabetes treatment decisions," and the record lists its generic name as "Sensor, glucose, invasive, non-adjunctive" [11]. The G5 still asked users for fingerstick values to calibrate, and FDA's database now shows that approval withdrawn as of March 20, 2026 [11].

Every iCGM indication we read carries the same idea. The G6, G7 and Libre Plus sensors are each intended to replace fingerstick or blood glucose testing for treatment decisions [2] [5] [6]. Abbott's wording adds four words that matter: "unless otherwise indicated" [6].

When the labeling names a meter

The exceptions are written out. The G7 summary says to use a meter when readings don't match symptoms, when the display lacks a number and a trend arrow, and during the 30-minute warm-up [5]. Abbott's 2024 summary lists a mismatch with symptoms or expectations, a missing number, and the "check blood glucose" symbol during the first 12 hours of wear [6]. The G6 summary went further, adding the period when someone is new to the device and times when glucose is rising or falling rapidly [2]. Which of these applies to you, and what to do with the meter result, is a decision to make with your care team.

Why meters stay the comparison method

In the pivotal studies, the reference is a lab analyzer, not a home meter. Even so, the G6 study compared its youngest group, children 2 to 5, against a blood glucose meter instead of a laboratory method [2]. Meters also remain the calibration tool when users want one: the G7 is factory calibrated, and users may enter optional fingerstick calibrations [5].

Clinical guidance hasn't dropped the meter either. Recommendation 7.10 of the ADA's 2026 Standards of Care says people using CGM devices "must also have access to BGM at all times" [12]. That is why a strip still has to meet its own 2020 accuracy targets: it is the check that CGM labeling names when the sensor and the person wearing it disagree.