CGM Accuracy
Accuracy, the Black Swan, and the Finger-Prick Habit
A nurse in clinic asks a question she has asked twenty times this year: when did you last do a finger-prick. The answer is honest, “I haven’t, the CGM is good now”. Most days that is right. Across two thousand readings on a single sensor wear, between twenty and forty of them will fall far enough from the truth to drive a wrong dosing decision. Those readings do not arrive labelled. The alarm is silent, the number looks like every other number, the response is the wrong response. CGM is one of the great safety-and-quality-of-life advances in T1D. The finger-prick habit is what protects against the small fraction it gets wrong.
Ask Grace
Want to ask when a finger-prick is the right call, why GNL does not lead with MARD, or what the ±40/40 outer band actually protects? Ask Grace.
Why GNL does not lead with MARD
The headline accuracy figure most manufacturers lead with is the Mean Absolute Relative Difference, the average percentage difference between the CGM and a reference. It tells you how the sensor performs on average. It does not tell you how often readings fall far enough from reality to drive an unsafe insulin dose. An average can look reassuring while hiding the tail; two sensors with the same MARD can behave very differently at the edges, and the edges are where dosing decisions go wrong. GNL therefore reports CGM accuracy using ±20/20 and ±40/40 agreement only. These are the numbers that describe the distribution of readings, including how often a device falls outside the clinically safe band. The DSNFUK CGM comparison chart uses the same measures, and so does every device page in this guide.
The accuracy zones
Three concentric bands tell you how confident to be in a reading: the inner zone where the CGM is close enough to dose insulin from, the middle band where it is still acceptable for trend reading but not for tight dosing, and the outer zone where the reading is far enough from the truth to be unsafe.
Two thousand readings, one sensor wear, twenty to forty in the outer zone. A 1% miss rate sounds small. On a single 10 to 14-day sensor wear at one reading every five minutes, it works out at 20 to 40 readings far enough from the truth to drive a wrong dose. The numbers do not arrive labelled. The alarm is silent. The CGM is confident. The finger-prick habit is what catches them.
The black swan, and the lost art of the finger-prick
Between 0.1% and 1.2% of CGM readings from framework-qualified devices fall outside the ±40/40 agreement band. This is not a design flaw unique to any manufacturer. It is a property of interstitial glucose sensing. Interstitial glucose lags blood glucose by 5 to 15 minutes; during rapid change, the lag compounds. Pressure on the sensor, hydration, exercise, and perfusion can all push readings temporarily outside the accuracy window.
This is not a risk-benefit question where the convenience of CGM is weighed against a small chance of error. The end consequence of acting on a significantly wrong reading in the wrong direction is severe hypoglycaemia or DKA. Those are not outcomes to trade off. The precautionary principle applies: when a small percentage of readings can produce a harmful outcome, the response is not to ignore them, it is to keep the ability to detect them.
Nobody is asking for a finger-prick on every reading. The 99% of readings inside the accuracy window represent a level of insight that was not available to people with T1D a decade ago, and the wear-day rhythm of CGM is one of the great quality-of-life advances in modern diabetes care. What the framework asks for is a working meter, in-date strips, and the habit of reaching for it when something does not feel right: the CGM says low but you do not feel low; the CGM says high but you only just bolused; an alarm fires but the clinical picture does not fit. The DSN Forum UK nurses who collaborated on this framework said it directly: CGM has pushed us a hundred steps forward, and from an education perspective we might need to take half a step back, to make sure every person using CGM has a working meter and the understanding that occasionally they will need it.
Accuracy inside an AID system
When a CGM is integrated with an automated insulin delivery system, the accuracy of the CGM has a direct effect on insulin delivery decisions made by the algorithm. A reading that drifts outside the safety band drives an insulin response that drifts with it, automatically, without a human decision step. This is why iCGM (integrated CGM) designation, required in the US for AID integration, demands stricter accuracy standards and tighter study design. The framework on the previous page applies to standalone CGM use; the same logic applies more strongly inside an AID loop. AID compatibility for each device is covered on the device pages.
Calibration, when it is on offer
Some CGMs accept an optional calibration: enter a finger-prick reading to refine the sensor’s accuracy, especially in the first hours of wear or when readings feel inconsistent. Calibration can improve accuracy when done correctly. It can introduce error when done badly, for example by entering a CGM reading rather than a true blood-glucose measurement (this is the circular calibration that offers no error correction; it has been flagged in clinic and in podcast Episode 36). Calibration availability is noted on each device page; SmartGuide has a mandatory day-one calibration sequence rather than an optional input.
Step 3 of 3
CGM Accuracy