Continuous glucose monitoring – how accurate are they?

13th April 2019, Dr Chee L Khoo

The monitoring of plasma glucose levels is crucial when we are titrating insulin doses, correlating lifestyle messages with glucose control and in detecting asymptomatic hypoglycaemia. Until now, finger-prick glucose checking is the gold standard for self-monitored blood glucose (SMBG). However, even with 7 point glucose monitoring, nocturnal hypoglycaemia may be difficult to detect. This is where continuous glucose monitoring (CGM) is in a class of its own.  Stand-alone CGM devices are increasingly being used to fine-tune the management of patients on insulin therapy. How accurate are these new devices, though?

CGM measure the glucose in the interstitial fluid using sensors placed under the skin. Glucose readings are checked every 5 mins and the data is transmitted wirelessly to a reading device which can include a smart phone. There are a number of standalone devices now available in Australia – Abbott Freestyle Libre®, Dexcom G5 Mobile® and Medtronic Guardian Connect®. The last two are now available under the NDSS in patients with T1D subject to a number of criteria. The Abbott Freestyle Libre (so-called flash glucose monitoring) is calibrated at the factory and do not need further calibration while the other two need finger prick capillary readings to calibrate the machines.

When we talk about “accuracy”, there are two things to consider: how consistent are the readings and how do the readings compare with the reference standard which in this case, venous glucose readings. Consistency is about how well the readings correlate with venous glucose readings. If the correlation is good, then what is the difference between the readings? This is often measured by mean average relative difference (MARD) and mean absolute difference (MAD).

In a small study involving 5 insulin treated patients with type 2 diabetes, interstitial glucose levels measured by Abbott FreeStyle Libre (FGM) and Metronic iPro-2 (CGM) were compared with capillary glucose (finger prick) and venous glucose readings. There were good correlations between interstitial and capillary readings with venous glucose readings.

In the glucose range between ~4.0 mmol/L and ~13.0 mmol/L, when compared with venous glucose readings, capillary glucose readings had a MARD of 7.2 +/- 7.2%. The MARD for FGM and CGM were 8.2+/- 5.6% and 9.2 +/- 9.1% respectively. The MAD were 0.57 mmol/L for capillary blood levels, 0.67 mmol/L for CGM and 0.63 mmol/L for FGM. In other words, the margin of error is about ~10% when compared with laboratory venous glucose readings in the glucose range of 4-13 mmol/L. We don’t know what the error range is outside those glucose ranges.

Although consistent, FGM readings was found to be generally lower when compared with venous readings.

The result in this small study was consistent with those previously reported by Bailey et al when they compared FGM readings with capillary readings (1). The readings were consistent and the MARD reported in that study was 11.4%. Aberer et al  (2) and Olafsdottir et al  (3) also found that FGM readings were 7.6% and 9.2% lower than venous readings in their respective studies. In yet another study, FGM was found to measure systematically lower readings than capillary readings.

The newer, second generation, ultra-long acting insulins is associated with lower incidence of hypoglycaemia because of their longer half-lives. This has allowed us to push the boundaries a bit further in the pursuit of tighter glycaemic control. Conventional finger prick SMBG may still miss nocturnal hypoglycaemia. We are increasingly relying on FGM or CGM to guide us with insulin dosing. We need to be aware of the “accuracy” of these devices and the possible underestimation of some of the devices.

A lower than normal reporting of glucose readings from FGM may warrant caution in interpreting the results of FGM. Nonetheless, a recent randomised study limited to patients with relatively good glycaemic control (HbA1c <7.5%) using FGM found a 38% reduction in patient time in hypoglycaemia compared with controls, indicating safety from a hypoglycaemia point of view. However, despite the lower accuracy of FGM, the more frequent checks by patients is leading to more time within glycaemic control and less time in hypoglycaemia. This translates to lower diabetes complications and safer patient outcomes.

Access the abstract here.


  1. Bailey T, Bode BW, Christiansen MP, et al. The performance and usability of a factory-calibrated flash glucose monitoring system. Diabetes Technol Ther 2015; 17: 787–794.
  2. Aberer F, Hajnsek M, Rumpler M, et al. Evaluation of subcutaneous glucose monitoring systems under routine environmental conditions in patients with type 1 diabetes.
  3. Diabetes Obes Metab 2017; 19: 1051–1055.
  4. Olafsd_ottir AF, Attvall S, Sandgren U, et al. A clinical trial of the accuracy and treatment experience of the flash glucose monitor FreeStyle Libre in adults with type 1 diabetes. Diabetes Technol Ther 2017; 19: 164–172.
  5. Tatsuya Sato, Hiroto Oshima, Kei Nakata, Yukishige Kimura, Toshiyuki Yano, Masato Furuhashi , Masaya Tanno, Takayuki Miki, Tetsuji Miura. Accuracy of flash glucose monitoring in insulin-treated patients with type 2 diabetes. J Diabetes Investig 2018.