What is Continuous Glucose Monitoring (CGM)?
A Continuous Glucose Monitor (CGM) is a wearable device that tracks glucose levels throughout the day and night, providing real-time readings with trend arrows and alerts for high and low glucose levels. CGMs are a critical tool for diabetes management, offering insights beyond traditional blood glucose monitoring (finger-pricking). The picture below explains what CGM is and how the trend arrows help to see into the future!
For the basics, read the CGM foundations.
However, not all CGMs are created equal, and choosing the right one requires more than just comparing advertised accuracy numbers along with the various bells and whistles!
Why Risk Assessment Comes First
Before comparing CGMs based on performance, we must first assess the risk of using a particular CGM. Think of it like scouting a football player.
If claimed I have scored as many goals as Messi and Ronaldo, yet, I cost half the price, would you sign me instantly?
Or would you first ask:
- Where did you score those goals? (Top leagues or Sunday league football?)
- How competitive were the matches? (Practice or real games?)
- How do you perform at the top level, in the biggest games, when it matters most, and under the greatest pressure?
A CGM might advertise a Mean Absolute Relative Difference (MARD) of 10%, which suggests that on average, the CGM readings are only 10% different from actual blood glucose levels. However, unless we know how and where it was tested, MARD and other performance statistics are, well… meaningless.
Without robust testing, choosing a CGM is like buying a car that has only been test-driven at the national speed safety limit in the middle lane, when in reality, people use all three lanes and sometimes change lanes rapidly, just as glucose levels fluctuate in real life.
Would you drive a car that’s only been tested like this?
The Three Key Questions for CGM Study Design
To truly understand a CGM’s performance, we must examine how its accuracy was tested. The POCT05 guidelines, the eCGM Clinician Consensus, and the IFCC Working Group for CGM outline three critical questions that determine whether a CGM has been properly assessed across real-world conditions:
What percentage of study participants had type 1 diabetes?
- If fewer than 70% of participants had type 1 diabetes, the CGM might not have been tested in conditions where glucose fluctuates rapidly.
- People with type 1 diabetes lack insulin production, meaning their glucose levels move faster than those with type 2 diabetes, allowing us to assess how well the CGM performs in extreme shifts.
Were meal and insulin challenges performed?
- A proper accuracy study should intentionally induce high and low glucose conditions by:
- Giving food without insulin to create rapid glucose spikes which happens in real-life from time to time.
- Giving too much insulin without food to induce fast glucose drops and hypoglycaemia – “Rage bolus!”
- Without these tests, the CGM may not have been exposed to the real extremes that people with diabetes experience from time to time.
What percentage of comparison readings were in extreme glucose ranges?
- A CGM must be tested in all glucose ranges:
- At least 8% of readings should be below 3.9 mmol/L (hypoglycaemia). Of course, people with diabetes do not generally have 8% hypoglycemia, but to check performance in this range, there must be adequate readings to assess performance.
- At least 5% of readings should be above 16.7 mmol/L (hyperglycaemia). Again, adequate readings are needed to assess accuracy in the very high range.
- If these extremes are not well-represented, the CGM’s ability to detect dangerous highs and lows is uncertain. We don’t want uncertainty about accuracy when treating hypos or giving large insulin corrections when very high.
Which CGMs Have Been Tested Properly?
Using these three key criteria as your basic minimum criteria, you can identify cars that have been tested in all three lanes going at different speeds.
Only the following CGMs meet the basic testing standards and have published (not data on file that has not been peer-reviewed!) their accuracy data for public review:
- Accu-Chek SmartGuide
- Dexcom G6 and One
- Dexcom G7 and One+ One
- EverSense
- Freestyle Libre 2 and 2 Plus
- FreeStyle Libre 3 and 3 Plus
- Medtronic Guardian 3
- Medtronic Guardian 4
- Medtronic Simplera
Any CGM that has not met these study design basics cannot be meaningfully evaluated, despite having CE marking that enables marketing across a broad age range with non-adjunctive indications!
More recently, CE marking for CGM systems driving “AID insulin dosing”, without peer reviewed published data, which has led to serious concerns raised by leading pediatric societies and adult specialist technology networks.
Why is this a concern?
We lack reliable data on CGM performance during the most critical therapy periods when glucose levels are in the low and very high ranges. This does not mean those CGMs are unsafe; it means we simply do not know. If you are in the business of taking unknown risks, knock yourself out.
What Should We Compare CGM Readings Against?
Now that we know which CGMs have been properly tested, the next step is understanding what we should compare CGM readings against.
CGMs can be compared to different comparators:
Venous Blood Glucose
- Drawn from veins (after glucose has been used by cells).
- Lower glucose concentration compared to capillary blood, by 5-10% overall, and even lower after eating when the glucose spikes!
Capillary Blood Glucose
- Drawn from small blood vessels (before glucose is absorbed by cells).
- Higher glucose concentration. This is the level your cells are exposed to, especially after eating, making it crucial for assessing potential damage and complication risk.
Key Takeaway
- Capillary glucose reflects the highest concentration of glucose your cells experience, making it a critical measure for understanding metabolic health.
Also, consider that a person with diabetes does not have an IV line in their veins at home to check the accuracy. They have a finger-prick tester to measure capillary blood glucose. Remember this: it is important to use a finger-pricking blood glucose tester that aligns with your CGM system. There will be more on this later once we know how closely CGM systems align with capillary blood glucose.
So, when assessing CGM performance, we must consider how the comparison method impacts accuracy. Let’s undertake a thought experiment.
The Holiday Budget That Didn’t Translate
Imagine you’ve been planning your dream holiday to the U.S. for years. You do your research thoroughly, checking the prices of hotels, restaurants, attractions, and everything else you want to experience. Everything is listed in U.S. Dollars ($), so you carefully budget, writing down the exact costs for each part of your trip.
But there’s one crucial mistake. You assume that 1 Euro (€) is equal to 1 U.S. Dollar ($). So, if you see a hotel that costs $200 per night, you save €200 per night, believing that when the time comes, your Euros will cover everything exactly as planned. You track all your savings against this false 1:1 exchange rate, never questioning it.
Then, the day arrives. You land in New York, excited and ready to enjoy your trip. But when you go to exchange your money, reality hits! 1 Euro is actually worth only 0.90-0.95 USD. That means all your careful planning was based on the wrong assumption, and suddenly, you’re 5-10% short on your entire budget. The hotels, meals, and experiences you thought you had covered now cost more than you expected. Some things you simply can’t afford anymore.
Application to Glucose Monitoring
This is exactly what happens with glucose monitoring today. Many CGM systems are aligned to venous glucose, just like you mistakenly planned your budget using Euros as if they were equivalent to U.S. Dollars. These systems assume venous glucose levels are the right reference point, and people tracking their glucose believe they stay in range.
But the real measure that matters is capillary glucose!
Why? Because,
What gets measured gets managed
Remember, capillary glucose is 5-10% higher overall and even higher after eating when compared to venous glucose. Just like your money didn’t stretch as far as you thought, people using venous-aligned CGMs may believe their glucose levels have been in range, only to discover later that their actual exposure to higher glucose levels was greater than they realised.
The long-term “cost”?
Very likely, a higher risk of complications, even though they thought they were doing everything right.
This means that 70% in the target range (3.9-10.0 mmol/L) for a system aligned to venous glucose leads to a higher HbA1c than the same 70% achieved on a system aligned at capillary glucose.
Want an example from the current research?
- Medtronic 780G users require approximately 75% TIR to achieve an HbA1c of around 7% (53 mmol/mol), or 80% TIR to reach 6.5% (48 mmol/mol).
- Control-IQ users need only about 68% TIR for an HbA1c of approximately 7% (53 mmol/mol), or 74% TIR for 6.5% (48 mmol/mol).
This discrepancy arises from differences in sensor calibration:
- The Medtronic 780G CGM aligns to venous glucose with a large negative bias to capillary glucose (the displayed CGM value for the Simplera is 17% lower than the capillary level), resulting in an underestimation of capillary glucose.
- The Dexcom G6 (used in Control-IQ) aligns to arterial-venous glucose (as does the G7), which aligns more closely with capillary glucose; the G7 only has a 3% negative bias.
- As a result, this data suggests you can get away with a 5-7% lower time in range (3.9-10.0 mmol/l or 70-180 mg/dL) for a Dexcom G6/G7 sensor for the same HbA1c as that achieved with the Medtronic G3/G4/Simplera.
Another real-life example to cement our understanding?
Imagine two students, Alex and Ben, both working hard towards their degrees. Each of them has a different professor grading their coursework.
Alex’s professor, Dr. Lenient, is a bit relaxed with grading. If an answer is close enough, it gets full credit. Ben’s professor, Dr. Precise, is much stricter, only awarding marks when answers fully meet the criteria. Despite these differences, both professors consistently score their students at 70%, making Alex and Ben believe they are on track for a first-class degree.
When their final dissertations are sent for independent adjudication, essentially the academic equivalent of an external HbA1c test, things change. The independent examiner follows a strict standard and re-evaluates their true level.
Alex, who was used to Dr. Lenient’s grading, discovers that some of his work wasn’t quite up to the expected standard. His final score gets downgraded, and he ends up with a 2:1.
Ben, who had already been scrutinised under a strict marking scheme, sees his grade hold steady, earning a first-class degree as expected.
Now, think of Alex’s professor as a glucose sensor measuring venous glucose, it’s systematically a little looser, underestimating the real performance. Meanwhile, Ben’s professor represents the sensor measuring capillary glucose, which is closer to the true metabolic reality.
So, even though both sensors (or professors) suggest the same 70% score, when reality (HbA1c) is assessed externally, the stricter standard (capillary glucose) aligns better with the true outcome, just like Ben’s first-class degree.
What does this mean?
This means we need to know how accurate a CGM system is compared to capillary glucose, not venous glucose.
And that is the next part of the guide,
Assessing CGM Accuracy Performance.
