Assessing CGM Accuracy Performance

Prefer a podcast series?

Listen to the CGM 3-Part Series.

A quick recap from Understanding Risk Before Performance.

First reminder: not all CGMs measure glucose in the same way. Some are calibrated closer to capillary glucose, others closer to venous, and some even sit below venous. That affects:

• How accurate they look on paper (MARD, 20/20, Error Grid).
• How much they under- or over-report true glucose exposure.
• What Time in Range (TIR) target you actually need (e.g. 70% vs 75–80%).

If you have not already, it is worth revisiting CGM Systems Guide – Part 1, which explains why different CGMs need different TIR targets because of calibration and comparator differences:

Part 1 – Why some CGMs need 5–10% higher TIR targets

Which CGM systems passed basic study design criteria?

Based on the DSN Forum UK CGM comparison work, the only CGM systems tested using study designs that meet the basic international criteria are:

• Accu-Chek SmartGuide
• Dexcom G6 and Dexcom One
• Dexcom G7 and Dexcom One+
• FreeStyle Libre 2 and Libre 3 (and Plus versions)
• Medtronic Guardian 4
• Medtronic Simplera

A CGM should aim to report glucose in the space where your tissues actually live — roughly between capillary and venous glucose, not below venous.

Capillary glucose represents the highest glucose levels that micro-vessels (eyes, kidneys, feet, sexual organs, nerves) are exposed to. This makes it the key driver of long-term complication risk.

In contrast, a CGM that systematically reads below venous glucose will significantly underestimate peak exposure, and will make your numbers and TIR look “better” than the body is really seeing.

This is exactly why different CGMs need different TIR targets and why their “accuracy” depends on what you compare them against.

In this three-sensor study, the Medtronic Simplera tends to under-report post-meal glucose exposure because its readings align more closely with venous glucose values. In contrast, the Dexcom G7 and FreeStyle Libre 3 track just below capillary glucose, which more accurately reflects the glucose levels the body is truly exposed to.

Devices aligned with physiological glucose (close to capillary)

These devices report glucose levels that align closely with capillary glucose and therefore reflect true physiological exposure. For these systems, the standard 70% Time in Range (3.9–10.0 mmol/L / 70–180 mg/dL) target is appropriate.

• Accu-Chek SmartGuide (Roche)
• FreeStyle Libre 2 (and Plus version)
• FreeStyle Libre 3 (and Plus version)
• Dexcom G6
• Dexcom G7
• Dexcom ONE
• Dexcom ONE+

---

Devices reading below venous glucose

These systems tend to read below venous glucose, meaning they under-report true glucose exposure. For many people, this means they may need around 75–80% TIR on these devices to achieve the same underlying exposure as approximately 70% TIR on the systems above.

• Medtronic Simplera / Simplera Sync
• Medtronic Guardian 4
• GlucoMen iCan
• ALLYcgm (AgaMatrix)
• TouchCare Nano A8 (Medtrum)
• Linx (Microtech)
• Gluconovo (Infinovo)
• GlucoRx Aidex
• GS1 CGM (SiBionics)
• Yuwell CT3
• Syai Tag (Syai Health Technology)

Key point repeated: because these sensors are calibrated differently, they do not just “display different numbers” — their accuracy metrics and your TIR target need to be interpreted differently too.

Performance metrics – why averages can mislead

There are lots of accuracy metrics, each with limitations. The most common is MARD (Mean Absolute/Relative Difference), which gives an overall average error. But an “average” tells you very little about risk at the extremes — where clinical decisions matter most.

Think of it like this:

“Would you cross a river if its average depth were 4 feet, even if you couldn’t swim?”

If that sounds safe, you may be underestimating risk. You need to know where the deep parts are, not just the mean.

CGM accuracy is similar. The extremes (lows and highs, and rapid changes) are where the most critical insulin decisions are made.

Agreement rates: 20/20 and 40/40

Agreement rates show the percentage of readings that fall within an acceptable error range, typically:

• 20/20 – “no clinical risk” zone.
• 40/40 – “low but non-zero clinical risk” zone.

20/20 metric – low-risk accuracy

A reading is within the 20/20 zone if:

• For CGM values < 5.5 mmol/L (100 mg/dL), the comparator is within ±1.1 mmol/L (±20 mg/dL).
  • Example: CGM = 4.5 mmol/L (80 mg/dL). The “true” value must be between 3.4–5.6 mmol/L (60–100 mg/dL).
• For CGM values ≥ 5.5 mmol/L (100 mg/dL), the comparator is within ±20%.
  • Example: CGM = 10.0 mmol/L (180 mg/dL). The reference must be 8.0–12.0 mmol/L (144–216 mg/dL).

The higher the % of readings within 20/20, the more dependable the CGM is for low-risk insulin dosing.

40/40 metric – detecting dangerous misses

The 40/40 metric is a wider safety net that helps flag readings with a high chance of leading to incorrect insulin decisions.

• For CGM values < 5.5 mmol/L, the comparator must be within ±2.2 mmol/L (±40 mg/dL).
  • Example: CGM = 4.5 mmol/L (80 mg/dL) → reference must be 2.3–6.7 mmol/L (20–120 mg/dL).
• For CGM values ≥ 5.5 mmol/L, the comparator must be within ±40%.
  • Example: CGM = 10.0 mmol/L (180 mg/dL) → reference must be 6.0–14.0 mmol/L (108–252 mg/dL).

If a CGM frequently fails even the 40/40 window, those readings are very likely to cause incorrect dosing decisions.

Important: because different CGMs are calibrated differently (capillary vs venous vs below venous), the same 20/20 or 40/40 percentage does not always mean the same real-world exposure. Again, calibration and comparator matter.

How do the CGM systems stack up?

The DSN Forum UK CGM comparison chart (published with permission) summarises study design scores and 20/20 / 40/40 agreement.

A note of caution when comparing systems

Tables like this are useful for checking whether CGMs meet minimum safety standards, but they are not true head-to-head comparisons. The underlying studies differ in:

• Populations (T1 vs T2, adults vs children).
• Protocols and glucose challenges (how hard the sensor is pushed).
• Comparators used (venous vs capillary).

In my view, if a CGM:

• Has a study design score of ≥ 4, and
• Achieves around 90% of readings within 20/20, and
• Has ≥ 99% within 40/40,

…then it can usually support low-risk insulin dosing decisions with minimal high-risk outliers.

But where calibration differs (venous-aligned vs capillary-aligned), two CGMs with similar 20/20 / 40/40 stats can still deliver different actual glucose exposure. That is the core calibration problem from Part 1, now showing up again in performance metrics.

iCGM standards, DGR testing, and why comparators matter

When a CGM system meets iCGM performance standards (which use venous glucose as the comparator), it is considered highly accurate and reliable. iCGM clearance is a strong reassurance.

However, when DGR tests (Dynamic Glucose Range challenges) use capillary glucose as the comparator, the criteria become extremely strict. Even top-performing CGMs such as Dexcom G7 and FreeStyle Libre 3 fail to meet iCGM criteria under those conditions.

This chart (orange and red bars) shows that even our best sensors “fail” iCGM when tested under a robust DGR protocol using capillary glucose as the comparator:

This does not mean the market-leading sensors have suddenly become less accurate. It simply highlights that:

• When you stress-test CGMs with DGR protocols and use capillary glucose as “truth”,
• The current iCGM performance thresholds become almost impossible to meet.

Again, the comparator (venous vs capillary) changes the apparent accuracy of the same sensor.

As DGR testing becomes standardised by the IFCC team, new performance criteria will be needed that reflect this more demanding and realistic assessment.

So, is it time for DGR-based performance standards?

I believe the case is strong.

We need metrics that test CGMs across the five critical DGR regions: BG Low, Alert Low, Neutral, Alert High, and BG High.

Why five regions?

This approach shows sensor accuracy when glucose is:

• Dropping towards hypoglycaemia.
• Rising rapidly into high ranges.
• Moving through the “neutral” zone where we make most fine-tuning decisions.

In other words, it reflects what users actually experience and where errors matter most.

Hint: we must be especially vigilant during rapid glucose changes, particularly drops triggered by exercise. This has been underlined by Professor Othmar Moser and the expert groups behind the international guidelines for CGM and exercise and AID and exercise in type 1 diabetes.

DGR metrics and standards are not ready yet, but they are coming.

The DTS Error Grid – visualising clinical risk