The GNL Podcast

Episode 20, Standardisation of CGM Performance Testing: The Nuts and Bolts

Two people wearing two different sensors stand in the same kitchen, glance at the same falling glucose, and read two different numbers off their phones. Both are about to make a decision about insulin. The question this episode sits on is a quiet one: how do we know either reading can be trusted, and what would it take to make them agree?

Ask Grace

Want to understand why two CGMs can report the same glucose yet produce different time-in-range values, and what an international standard would do about it?

Listen

Listen on Apple Podcasts Listen on Spotify Listen on Buzzsprout

Also available on YouTube. Guest: Dr Stefan Pleus, Institut fuer Diabetes-Technologie, Universitaet Ulm, and Chair of the IFCC Working Group on CGM. Host: John Pemberton. Director of Creativity: Anjanee Kohli.

Why this episode exists

If you live with type 1 diabetes, you have probably noticed that two CGM systems can report the same glucose level and yet hand you very different time-in-range values, and different implied HbA1c readings. You may have assumed the numbers were interchangeable. They are not, and that inconsistency quietly undermines trust, therapy adjustments, and clinical trial endpoints. The reading on the screen is only as good as the testing that stands behind it, and right now that testing is not standardised across devices and manufacturers.

This episode brings in Dr Stefan Pleus of the Institut fuer Diabetes-Technologie, Universitaet Ulm, and Chair of the IFCC Working Group on CGM, to explain why the current FDA iCGM framework (2018) is no longer sufficient, what a robust international standard would look like, and why ISO standardisation by 2030 is essential. It is the standards-and-regulation companion to the broader CGM accuracy work on GNL, and follows directly from Episode 19 with Dr Guido Freckmann, which covers why current frameworks fall short.

In this episode

Dr Stefan Pleus explains what rigorous CGM performance testing actually requires, and what it will take to get there. The conversation moves from the limits of today’s testing to the shape of a robust international standard, and to the practical question of why ISO standardisation by 2030 matters for everyone who relies on a sensor reading.

The thread running through the discussion is consistency. Pleus sets out why ISO standardisation by 2030 is essential if CGM is to serve as a reliable basis for insulin dosing decisions, act as a trusted comparator in clinical trials, support screening and early diagnosis of type 1 diabetes, and provide people with diabetes consistency across devices and manufacturers.

What this episode explores

Why CGM should serve as a reliable basis for insulin dosing decisions
How CGM can act as a trusted comparator in clinical trials
How standardised CGM could support screening and early diagnosis of type 1 diabetes
How a standard would provide people with diabetes consistency across devices and manufacturers

Watch or listen

Key themes

Easy study conditions flatter every sensor

A study can be designed to look good or designed to be honest. As Pleus puts it: “If you design your study material and methods to avoid rapid glucose changes, you will get good results, but that’s not real life. Many systems perform worse at high rates of change, and unless you test that, you don’t know the risk.” The conditions a sensor is least likely to handle well, rapid change, are precisely the conditions everyday users live in.

Stress-test zones, not just the comfortable middle

The IFCC Working Group’s answer is to insist the testing covers the difficult parts of the range, not only the parts that are easy to measure. In Pleus’s words: “We introduced comparative data characteristics, requiring minimum data in low, high, and rapid change zones, so that stress test conditions are covered, not just the easy ones.” A standard is only useful if it forces the test into the conditions that carry the most risk.

You cannot align readings without correcting the comparator

Aligning CGM readings across manufacturers depends on the reference method being sound, not just the sensor. Pleus is direct about it: “Every method has bias. Unless you retrospectively correct comparator results against higher-order reference materials and methods, you cannot align CGM readings across manufacturers.” Without that correction step, comparing one device’s accuracy claim to another’s is comparing two differently calibrated rulers.

Today’s devices will look less accurate, and that is the point

A more demanding standard does not make sensors worse; it stops flattering them. Pleus frames the consequence plainly: “With a more robust study design, today’s systems will look less accurate, not because they’ve changed, but because we’re finally stress-testing them properly.” The drop in the headline figure is the standard doing its job, not the device failing.

The barrier is regulatory will, not the science

An international standard typically takes 3 to 5 years to develop, and the slow part is not the technical work. Pleus is clear about where the friction sits: “An international standard typically takes 3-5 years. Without alignment between FDA and Europe, we risk duplication, cost, and stalled innovation. The biggest barrier is regulatory will, not the science.”

The headline reassures; the testing decides what it is worth. Different CGM systems can report the same glucose level yet produce very different time-in-range values, and different implied HbA1c readings. Without standardisation, clinical trials, diagnostics, and everyday insulin dosing risk being inconsistent across devices. The opportunity, by 2030, is a global, reproducible framework that ensures fairness for users, clarity for clinicians, and valid results for researchers.

Practical exploration

For people living with type 1 diabetes and their families

The takeaway is not a rule to follow; it is a way of reading the numbers in front of you.

If you have used more than one CGM, notice that the same glucose level can produce different time-in-range values and different implied HbA1c readings between devices; the difference does not mean one device is lying, it reflects the absence of a shared testing standard.
When you read an accuracy claim, consider whether the testing behind it included rapid glucose change, or only stable conditions, because many systems perform worse at high rates of change.
Treat a standardised, reproducible framework as something worth advocating for, since it is what would let you move between devices and manufacturers with more consistency.

For clinicians and educators

The standards conversation belongs in the same critical-appraisal frame you already apply to clinical evidence.

Recognise that inconsistency across devices undermines trust, therapy adjustments, and clinical trial endpoints, and factor that into how you compare metrics between sensors.
When CGM is used as a comparator in trials, or to support screening and early diagnosis of type 1 diabetes, the absence of a standard is a methodological consideration, not a detail.
Follow the IFCC Working Group on CGM publications and the Diabetes Care commentaries below to track how the standardisation case is being built and contested.

About the guest

Dr Stefan Pleus is based at the Institut fuer Diabetes-Technologie, Universitaet Ulm, and is Chair of the IFCC Working Group on CGM. He works on the design and rigour of CGM performance testing, and on the case for an international standard that would make sensor accuracy claims comparable across devices and manufacturers.