CGM Framework
The CGM Framework
A meeting room at the DSN Forum UK in late 2023. Three diabetes specialist nurses going through the manuals of every CGM they were being asked to recommend to children in their care. Two devices carried a CE mark and an indication for use in children from age two. Neither had paediatric data in the studies behind that indication. The CE mark said yes; the data did not say the same thing. The framework on this page came out of that room. It is not a feature checklist; it is a five-question filter that asks whether the accuracy data behind a CGM is good enough to dose insulin from.
Ask Grace
Want to ask why a particular CGM scores 5/5, or what changed between the original DSNFUK chart and the framework on this page? Ask Grace.
Why a CE mark is not the same as evidence for insulin dosing
Regulatory approval is a market-authorisation standard. It assesses whether a device is safe enough to sell. With blood glucose meters, ISO 15197 sets the clinical-quality bar for insulin dosing accuracy on top of approval. With CGM, no equivalent ISO bar exists for study design, study population, or how much data is enough. Between 2018 and 2022, several CGMs entered the market with an indication for use in children carrying accuracy data only from adults with type 2 diabetes. The indication said one thing; the studies behind it said another. This framework fills that gap.
Five questions, every CGM must answer all five. The framework is a floor, not a ceiling. A 5/5 score means the accuracy data covers all five clinical situations where a low-quality answer creates real risk. Anything below 5 means there is at least one situation where the evidence behind the device is insufficient to support dosing decisions with full confidence.
The five criteria, in detail
01, Were people with type 1 diabetes included?
Insulin physiology, glycaemic variability, and the risk profile of dosing errors differ between type 1 and type 2 diabetes. A device validated only in type 2 cannot be assumed to perform the same way in T1D, where insulin is the primary treatment and dosing errors carry greater consequences.
02, Was a meal challenge included?
CGM readings change fastest in the postprandial period. This is also when insulin dosing decisions are most frequent. A study without a standardised meal challenge cannot adequately assess accuracy at the moment that matters most.
03, Was a hypoglycaemia challenge included?
The consequences of a CGM reading that is wrong when glucose is genuinely low are severe. A study without a controlled hypoglycaemia challenge cannot tell you how the device performs at the levels where a dosing error is most dangerous.
04, Was a venous or capillary reference used?
Accuracy is meaningless without a reference. CGM measures interstitial glucose, not blood glucose. A well-designed study compares the CGM reading to a simultaneous venous or capillary blood-glucose reference under controlled conditions. Studies that compare a CGM to itself or to uncontrolled comparators introduce circular errors.
05, Were the population and indication matched?
An indication for insulin dosing in children from age two requires data in children. An indication covering pregnancy requires data in pregnancy. If the claimed indication and the study population do not match, the approval is built on assumption, not evidence.
How the framework is applied
Each criterion is scored as met or not met. A 5/5 score means the device’s accuracy data adequately covers all five risk areas. The score is a minimum standard, not a full clinical endorsement; a device that scores 5/5 may still have limitations in specific populations or contexts. Scoring is applied by John Pemberton in collaboration with the DSN Forum UK using publicly available accuracy data and regulatory documents, and updated when new evidence is published. Manufacturer input is invited to verify factual detail on device pages; it does not influence the framework assessment itself.
Step 1 of 3
The CGM Framework