Since going on a low carb diet last November, I have been keeping a close watch on my BG using my trusty OneTouch Ultra which I’ve had for years. Because it has consistently matched the BG values from lab tests, I’ve taken it as trustworthy but the discussions on the forums about meter accuracy made me want to find out a bit more and do some testing for myself.
I asked some companies about the data on accuracy of their meters and most didn’t reply but one company, DiME who market the GlucoRx, got in touch with me. The company not only provided me with test details and some documents but also sent me their new Nexus meter to try out along with a supply of test strips.
I have already posted a blog article which goes into considerable detail about meters and their accuracy and it should be understood that all meters have the same technical limitations including the OneTouch. This comparison is necessarily somewhat impressionistic and as with all such informal trials, there could be many reasons for a discrepancy. All meters on the market are tested to a high standard but the details of their clinical trials are not often available to the public – perhaps if they were, people like me wouldn’t try to test them for ourselves. The Nexus guys have provided details which show a high level of accuracy in their clinical testing; what follows is my personal experience.
I had already tried the previous GlucoRx which I found to be highly variable to the extent that I didn’t really trust it. Either I was not using it correctly, or the strips I had weren’t reliable, or the meter was faulty, or it was one of the many other sources of error. Whatever the situation, I had my concerns about the GlucoRx and how I was using it. I started finding out about the technology employed and the standards in force. Then, given the very positive and supportive attitude of DiME towards what I was trying to do, I agreed to try out the new Nexus. The Nexus is based on greatly improved technology using a more sensitive enzyme system. Below is a summary of what I did and what I found.
I am not posting the full data (on space grounds) but if anyone wants a copy, they can PM me and I’m happy to provide them.
This is not a clinical trial, nor any kind of controlled, randomised study so the results can and probably should be questioned on methodological grounds; it simply represents the results of my own personal testing. But I did try to follow instructions and my experience is likely to be fairly typical. I did though fail to discard the first drop of blood which might have affected the results.
A comparison was made of the values obtained by simultaneous tests of blood glucose using the OneTouch Ultra and the GlucoRx Nexus over a period of thirty-three days. The OneTouch has been compared to laboratory results on four occasions and has always been within 0.2 mmol/l of the laboratory reading.
Finger pricks were taken several times per day to a total of 88 in each case using the same hole to produce the blood sample for each meter. The first meter to be tested was alternated but no record was kept of which meter was tested first. Prior to every test, the hands were thoroughly washed and dried. Different fingers were used each day. Dates and times were recorded at the time of the test in every case. No values were discarded.
On one occasion, on 28th March, the tests were carried out in a clinic during a routine blood test and in that case, the drop of blood taken was from the vein following removal of the needle. Both meters were prepared with a strip just before the needle was removed.
Although the fingers were clean and the surface was lightly rubbed to ensure adequate circulation prior to pricking, I did not discard the first drop. Only later did I realise that this is recommended in the case of the Nexus. One reason for doing so may be that the first drop can contain cell debris and also in the case of too small a hole, the squeezing can affect the concentration of blood cells (the haematocrit) which can in turn affect the detected BG concentration.
Of the 88 values recorded for both meters, the Nexus was higher in 74 cases, and lower in only 14. The maximum difference was when the Nexus recorded 2.8 mmol/l higher than the OT, and the minimum difference was 0.1 mmol/l. A total of fifty values were within 1.0 mmol/l, 57%.
The average percentage difference between the Nexus and One Touch values was 11.24%, the Nexus consistently being the higher value. The Nexus was at least 10% higher in 53 cases, 60% of the recorded values. Where the OT value was 6.0 or less, the average difference shown by the Nexus was 0.6 mmol/l higher.
On the occasion where the values were compared to a laboratory test, the Nexus recorded 8.9 mmol/l and the OneTouch recorded 6.6 mmol/l. This was a surprisingly large difference so a second strip was tested on the Nexus and it gave a value of 8.7 mmol/l. The laboratory figure came back as 6.7 mmol/l. There was no apparent reason for the discrepancy. Since venous blood is typically 10-15% higher in BG than capillary blood, I would have expected a higher value so it is quite possible that the OneTouch recorded a value which was too low. In retrospect, I should have retested both meters and not just the Nexus. Given the rigorous standards of testing of all meters, it is far more likely that this discrepancy was the result of poor technique than any response from either meter. Results like these indicate the importance of controlled clinical trials data in making judgements about accuracy.
For diabetics the most important information is about whether the glucose level represents a high or a low value, rather than the exact measurement. The reason is simply that the diabetic has to take action based on that judgment so to some degree the exact value matters less. If there is inaccuracy, what has to be evaluated is how the patient might behave differently. If a value is judged to be high, how will the patient respond differently from a lower value? Will they take additional medication, change their diet, exercise more, and will the reading cause them anxiety and worry? And what will be the consequence of the action they take?
The results of my testing show that for the most part, the Nexus gives a general indication of the BG level which compares reasonably with the more expensive One Touch meter. Although the Nexus seems consistently to show higher BG levels compared with the OneTouch, the clinical trials data provided for the Nexus itself shows a high degree of accuracy. We cannot conclude solely on the basis of my informal trial that the Nexus is any more or less accurate than any other meter – for that, we need controlled clinical trials for all of the meters.
What might explain one meter consistently showing a higher level than another meter? There are countless reasons explored in my article BG Metering – A Realistic View. We have to consider the results of my informal trial in the context of all of those factors which affect BG readings.
An important question is whether the difference in reading is likely to result in significantly different patient behaviour. The higher readings I obtained may have been down to not discarding the first drop though it is not clear to me how this would produce a consistently elevated set of readings, especially since at times, each meter was used first. The Nexus makes use of a glucose dehydrogenase enzyme which, in contrast to other dehydrogenase-based meters, does not react to sugars other than glucose, which removes one source of variability. This should make it more accurate overall. In addition, there is the clinical trials data which shows a high level of accuracy. I don’t have an explanation of why the recorded levels were consistently higher but this is only an informal trial, not a laboratory controlled trial. I wouldn’t at all claim that my data is more reliable, but it is the data that I obtained on a daily basis and which informs my treatment choices and in that respect it represents likely data obtained from common practice.
The Nexus is much cheaper to use than the OneTouch Ultra since the OTU strips are currently around £22 for 50, whereas the Nexus strips are retailing at under £10. At a time when many diabetics are having their strips rationed by GPs, that has to be an important consideration.
Although the Nexus seems during my trial to register higher readings, the risk of having a hypoglycaemic episode because the BG level may be overestimated and the patient therefore medicates to correct it seems remote since at lower BG levels, the differences between all meters are also very small. All meters have the same limitations in that at lower ranges the absolute value of the likely error is small, but at higher levels the error can become more significant clinically.
At lower BG levels, the difference between the meters is much reduced but this is also precisely the range where patients already have the condition well-controlled. At higher ranges, particularly over 8.0 mmol/l diabetics, especially Type 1 diabetics on insulin and also those trying to follow a low carbohydrate diet, will be carefully evaluating the effect of different foods on their levels and their medication needs. Inaccuracy in this range can have a significant impact on patients’ dietary choices. There may be occasions when a particular food is judged to have a greater impact than in fact it has, but the consequent behaviour of the patient will often be in a beneficial direction even if a meter overestimates the BG value. But for Type 1 diabetics meter errors, which affect all meters, can have a serious effect on the risk of hypo.
The general level of inaccuracy in BG meters is a constant concern for diabetics but this is partly because of a confusion between precision and accuracy. Although both virtually all BG meters display BG to one decimal place, the stated accuracy standard is much less. Patients therefore assume that a reading of 6.5 indicates somewhere between 6.45 and 6.54. The reality is that it could be somewhere between 5.2 and 7.8 and that is significant. Any Type 1 diabetic on insulin, or any patient following a low-carb diet, would make very different choices depending on the which level was recorded.
Having said that, we should recognise that many meters perform substantially better than the minimum standard required. One study showed that the best meters had 92% of their readings within 10% and 98% within 15% of the true value. But although many meters perform at higher levels of accuracy, we need to understand where the errors are most common.
What patients on low carb diets therefore need is the information about the distribution of errors recorded from a controlled trial. If they can be confident that the errors within the ranges they experience are relatively low in absolute terms and that they occur infrequently, for example that in the range 5.0 to 8.0 90% of errors are within 0.4 mmol/l but that errors occur in only 5% of readings, that would enable them to make realistic use of the values recorded.
On the other hand, simply assuming that there may be up to 20% error, although that is the current EU expectation, does not reflect the real performance of their meters – their performance is actually much better within the significant ranges. Patients do not get sufficient information about the range and distribution of errors on their meters but this could provide manufacturers with a significant increase in trust in the marketplace.
For diabetics, some indication of the expected absolute error within particular ranges is much more useful than a generic 20% error rate. Perhaps manufacturers may not want to issue this information but to do so would help diabetics in the practical management of their condition and also increase trust in the use of the meters.
It is somewhat ironic that whilst manufacturers are trying to improve the technology, governments are reducing the availability of test strips on prescription. Currently many meters have strips that are too expensive for many people to carry out the level of daily testing which would enable them to properly control the condition using low carbohydrate dieting. There is pressure on diabetics from GPs to make less use of testing strips to save prescription costs and that, coupled with patient concerns about accuracy, is a worrying trend.
The Nexus comes in at less than half the price for strips compared to the OneTouch Ultra and as long as we understand the general points about the accuracy of all such meters, it looks promising. Reducing the cost of the strips helps of course, but there is also a need to improve confidence in the trustworthiness of the meters themselves. Manufacturers can do a lot to improve the way meters are presented to the diabetic community by making available more information about the distribution and absolute size of errors together with some scenarios interpreting actual readings.