Glucose monitoring system accuracy – what’s in a number?
When you see a glucose number on a screen, it’s accurate, right? Most people would say, ‘Of course, what’s the point otherwise?’
BUT for several reasons, blood glucose meter accuracy is less of a yes-or-no question and more of an ‘it depends’ scenario. This article covers the main factors that can affect glucose monitoring system accuracy and tips for ensuring you’re getting the best quality information to inform your diabetes treatment decisions.
Fingerstick blood glucose (BG) vs. interstitial fluid (ISF) glucose monitoring
Traditional fingerstick blood glucose meters measure glucose in capillary blood. Flash glucose monitoring (FreeStyle Libre) and continuous glucose monitoring (CGM, e.g. Dexcom, Medtronic, Eversense, Medtrum) devices are usually worn on the abdomen, upper buttock or back of the arm – depending on the device – where they measure glucose in interstitial fluid (ISF).
Capillary blood and ISF have different physiological characteristics. Changes in glucose levels will be seen more quickly in capillary blood (blood glucose, or BG). This phenomenon is often called ‘lag’. Interestingly, the lag between ISF glucose values and BG can vary between individuals, apart from the specific ISF glucose system they use. Algorithms in ISF glucose systems attempt to ‘smooth out’ differences between the glucose values they detect and the expected BG value.
Accuracy in BG monitoring
In 2013, the International Organisation for Standardization published revised quality standards for blood glucose meters in order to improve accuracy and consistency of results. Following the transition period, the UK adopted this standard as EN ISO 15197:2015.
ISO 15197 has 4 significant improvements on the previous standard (published in 2003).
- Increased accuracy for glucose meter systems, in particular for glucose values equal to or greater than 5.5 mmol/L
- Manufacturers of glucose meter systems must ensure their technology enables accuracy to improve from ±20% to ±15%
- The current standard covers 99% of results, as opposed to 95% under the old one
- For the first time, the standard provides formal acceptance criteria for accuracy as regards testing by patients as well as assessment of potential interfering substances (including haematocrit)
What does ±15% mean in practice? For example, when a laboratory test would say 6.0 mmol/L, a blood glucose meter that meets the ISO standard could show anywhere between 5.1 mmol/L and 7.0 mmols/L and still be called accurate. For more examples of how accuracy standards work in real life, Diabetes.co.uk has a good reference table (second table on the page) here.
Accuracy in ISF glucose monitoring
ISF glucose monitoring device manufacturers often use mean absolute relative difference (MARD) to describe performance. MARD for an ISF glucose monitoring device is calculated quite different from BG meter accuracy. It’s not possible to say a BG meter and an ISF glucose monitoring device are “equally accurate” because their relative accuracy is assessed differently.
For an analogy: the height of a building depends on the height of each story and the pitch of the roof. For this reason, not all 5-storey buildings are the same height. The ‘accuracy’ of an ISF device depends on its sensor chemistry, the algorithm it uses, its calibration, and the wearer’s body chemistry. It’s tempting to compare a CGM reading with a BG meter reading to get a sense of whether the CGM is ‘right.’ If the numbers are pretty close, good stuff; if they are very far apart, it’s probably best to rely on the fingerstick meter result at that point in time.
Aren’t all BG meters accurate though?
It’s a surprise to most people, including doctors and nurses, that a blood glucose meter doesn’t have to be independently assessed to be placed on the market in the EU, including the UK. The CE Mark process for BG monitoring systems relies on the manufacturer’s own assessments of accuracy being checked over by a Notified Body. It is possible for a manufacturer to choose the data they will submit for this assessment. So a product can get a CE Mark, be sold on the market, its test strips can be prescribed on the NHS – and it might not always perform according to the data submitted to obtain regulatory clearance.
In 2015, the Diabetes Technology Society, based in the USA, independently tested 18 popular blood glucose monitoring systems. They found some very interesting results, which you can review here. Several of the systems they tested are also available in the UK; some of them aren’t. A system that isn’t sold in the US might be highly accurate, it just won’t appear in the Diabetes Technology Society’s data set because it wasn’t possible for them to evaluate it. Still, this resource shows how performance can vary among marketed blood glucose monitoring systems.
What can I do?
- Get the foundation right: clean your hands before you do a BG check. Ideally, wash with soap and water; in a pinch, consider using a DiaWipe. Making sure there’s no food or drink residue on your hands is essential for an accurate result. If you haven’t washed your hands and the result would lead you to take a correction dose of insulin, clean your hands and then recheck first.*
- Consider your options in BG monitoring. Read up on different types of meters and have a conversation with your diabetes team and/or GP about what you think would work well for you. We have a guide here to what NICE says about choice in BG monitoring to help with this.
- Store and use test strips, sensors, meters and/or receivers in line with the product labelling (user guides and/or what’s printed on the test strip pack). BG and ISF systems depend on enzymes and to measure glucose levels. If a component has gotten too hot, too cold, too old, or damp, it might not work properly and could give a less accurate result.
- Keep calm and play it safe. A BG meter can give a somewhat different result if you check your BG again a few moments later. This is particularly true if your glucose level is quite high. If you tend to have a hypo after correcting a high blood sugar, it may help to correct to a more conservative target and then fine-tune from there. For example, if your regular ratio is 1 unit of insulin to 2.5 mmol/L, your target is 6.0 mmol/L and your current glucose is 14.3, the recommended correction would be 3.3 units. (If you’re on injections, you’d probably round this down to 3 units.) However, bearing in mind that a 14.3 result on your meter represents a potential range from about 11 mmol/L to 17.5 mmol/L, you might want to raise your correction target to 8.0 mmol/L and see where you land. This would mean taking 2.5 units instead. (If you are on MDI, it’s worth having a rapid-acting pen that can do half-units so you can make these types of adjustments – ask your GP or DSN about a half-unit pen for rapid-acting insulin if yours only does whole units.)
*Once I had a result of 16.5 mmol/L. I took a few units of insulin, washed my hands, then rechecked my glucose level. It was only 7.2 mmol/L! I drank three glasses of orange juice to avoid a crashing hypo. Not a great moment.