To start with, there are certain indications that blood glucose meters (BGMs) constantly show systematic bias, which is typically evident in lower glucose level readings compared to laboratory values. I was inspired by Chris Hannemann’s independent research, where Chris provided a comparison of BGMs based on finger blood sample.

In contrast to the finger prick blood sample tests provided by Chris, this research goes much further to a more accurate level using standard BG laboratory tests. It is focused on the comparison between glucose meter reading and laboratory value using the same vein blood sample. The main goal is to find glucose meters on the market that are as accurate as possible, with the lowest bias compared to the laboratory value. The other goal is to prove the suspicion of the systematic bias in BGM readings.

The results will help the T1D and T2D (diabetes type 1 and type 2) community to manage the diabetes including the HbA1C targets. The knowledge of the systematic bias can be beneficial in order to understand how the specific BGMs measure the glucose level.

Initial status quo

1. All the BGMs used in this research comply with the well known ISO norm 15197 no matter what release date is. Details about the ISO norm are not important in this research.
2. The blood glucose laboratory analysis is the most accurate available method of measuring BG levels, so it was used as a baseline to compare vein blood sample with BG meters.

BGM set for testing

Before the research was performed members of Facebook diabetes groups were surveyed about the most used BGMs. As a result of that questionnaire the following set of BGMs was put together:

Manufacturer	Model
Abbott	FreeStyle InsuLinx
Abbott	FreeStyle Libre
Abbott	FreeStyle Optimum Neo
Ascensia	Contour Next ONE
Bayer	Contour Plus
FORA	Diamond MINI Bluetooth DM30b
FORA	Diamond PRIMA DM10 Black
FORA	Diamond PRIMA DM10 White
LifeScan	OneTouch Select Plus
LifeScan	OneTouch Verio
Roche	Accu-Chek Performa Nano
VPD	2in1 Micro
Wellion	Calla Light
Wellion	Calla Mini

 

The Czech and the Slovak BGM distributors were kind enough to provide free samples of BGMs.

Test procedure

How the BGMs were tested?

The following procedure was set and accepted by a laboratory specialist.

1. Blood sample was taken in a laboratory only when the CGM readings were stable (i.e. not rapidly decreasing or increasing).
2. The reasonable representative set of glucose, from lows to highs, was taken to purposely reach different BG meters precision levels.
3. The Blood samples were kept in specimen tube designated for preserving blood.
4. A BG meter test was conducted during the hour after the blood was taken to avoid blood degradation.
5. Test strips were taken from vials just before the test and inserted into the test slots to activate the BGMs.
6. Blood was taken in with a syringe from a specimen tube and a blood drop was applied to the test strips.

 

The testing was made possible with the help of the Czech laboratories SynLab and Aesculab.

http://www.synlab.cz
http://www.aeskulab.cz

Data collection

Test data was filled into Microsoft Excel sheet to record values like “Test ID, Test Date / Time, Blood Sample ID, Meter Model Meter, Serial Number, Test Strip Lot #, Test Strip Expiration Date, Result [mmol/L]”. There are 200 test records in total. Bias from Lab  [mmol/L] was calculated from laboratory and BG test measurements. The bias is the most important value in this research.

Analysis and results

The measurements were analyzed in Microsoft Excel using standard descriptive statistics.

The following chart is the main one showing the difference between BGM readings and laboratory measurement. The baseline is the laboratory value. The X-axis shows the test identification number (a blood sample) while the Y-axis shows the bias in mmol/L.

Next is the interpretation of the descriptive statistics measures.

 Data variability

From this standpoint of the data variability we can see that BGMs from manufacturers Foracare, Roche and VPD have the least deviation from the mean.

 

Similarly the range here confirms Fora Diamonds, Roche and VPD as the most acceptable BGMs for accurate readings.

 

Data accuracy

The variability of BG readings is very high for Contour Next One and Wellion’s. The most accurate from this standpoint are Fora Diamond’s and Roche Accu-Check.

Similarly to the confidence level even standard error confirms Fora Diamonds and Roche Accu-Chek as the most accurate BGMs.

Skewness, a measure of asymmetry, tells us here what BG readings presents in the normal, symetric distribution. Let’s assume Contour Next One, Fora Diamonds and Roche are the most precise BGMs from this measure standpoint.

 

Ambiguities in results

Wellion

It looks like the unprecedented peaks of the bias were caused by unstable enzymes in test strips. Test strips were used according to instructions but still it seems after opening the vial, enzymes degraded quickly.

Lifescan

Lifescan completely changed test strip technology some time ago. We can see very different systematic bias since then. OneTouch Select Plus uses older technology while Verio newer one.

VPD 2in1 Mini and Ascensia Contour Next One

Both these BGMs were obtained at the end of research, so the results are affected by lack of test samples. Lets consider it in final resolution.

Here are the champions!

Most of the quality indicators meet following BGMs.

1. Fora Diamond Prima DM10
2. Fora Diamond Mini DM30b
3. Roche Accu-Chek Performa Nano

 

Photos of the winners

 

What to do next

To raise results onto a statistically significant level hundreds or thousands of tests must be done. One way to do this on larger scale project could be to distribute the work on volunteers all over the world and consolidate data into large database of BG measurements. If any of you are interested in joining, please send a message to “research at gandalf.cz”.

Conclusion

Why are there such differences between laboratory values and BG meters? One reason is definitely the technology used in BG meters/test strips. Second reason, and might be the most important, is the manufacturer – factory calibration. There were discussions with people from two different manufacturers. They claimed that the system bias is specifically set due possible law suites from people who were in hypoglycemic danger. Means that BG meters shows lower values for prophylactic reasons.

So even if the results are a bit affected by a small scale research, it is obvious that the original hypothesis about systematic bias was proven.

Disclaimer

Please consider this research was small scale. It was performed on personal basis so there were certainly budget and resource limits.

Contact author

In case you are interested in detail data feel free to contact author via email “research at gandalf.cz” or following contact form.