It’s impossible to manage your blood sugar well if you don’t know what your blood sugar level is. People with diabetes – both type 1 and type 2 – rely utterly on our measurement devices to make important treatment decisions and to refine our food and exercise decisions.
Blood sugar meters (“fingersticks”) and continuous glucose monitors both give exact measurements. But are the devices actually that precise? What’s the margin of error? And are some devices more accurate than others?
Glucose Meter Accuracy Standards
Here’s the FDA’s nonbinding recommendation for fingerstick glucose meters sold in the United States. These devices should deliver measurements that are:
Within +/- 15 percent of a highly precise lab measurement, 95 percent of the time
Within +/- 20 percent of a highly precise lab measurement, 99 percent of the time
So, if your blood sugar meter tells you that you’re at 120 mg/dL, most of the time that means that you could really be anywhere between 102 and 138.
The higher your blood sugar, the bigger the range gets. If your meter tells you you’re at 200 mg/dL, you may actually be anywhere from 170 to 230, a span of 60 mg/dL. That’s quite a range, and the device seems even less precise when you consider that as many as one in twenty measurements will fall outside of that range.
This spread of acceptable values catches some people by surprise. Many of us have taken two consecutive fingersticks, or used two different devices at the same time, and been shocked at the discrepancy between the two measurements. But quite often these wildly divergent measurements are safely within these established tolerance standards.
By the way, it is also difficult for these meters to measure very low and very high blood sugar levels with any accuracy, which is why many meters will show readings below 20 mg/dL as “LOW” and above 500 mg/dL as “HIGH.” Certain meters may have different ranges before these error warnings kick in – it may be worth checking the documentation on yours.
Glucose Meter Performance
There can still be quite a lot of variation between devices within the above standards. Some glucose meters are more accurate than others.
Unfortunately, it’s difficult to find reliable evidence on performance. Individual manufacturers might share data in the user’s manual, but this can be difficult to interpret. It also may not really apply to the device in your hands. The FDA has no mechanism to review glucose meters after they’ve already been released, and although they do require some continued testing of new lots of test strips, there isn’t really a good way to be sure that devices are as accurate as they day they were approved.
There’s quite a bit of evidence that many of these devices do in fact fall short of industry standards. This 2017 study compared 17 commercially available glucose meters, and this 2018 study evaluated 18. The trials had similar results:
Some devices were found to be much more accurate than others
A majority of the systems tested failed to satisfy established accuracy criteria
All of this data is very old – old enough that it may not be worthwhile to name the exact devices that were rated as most accurate. If you’re interested, you can click on the links above.
Other factors can negatively impact accuracy, too, including dirty fingers, test strip condition, temperature and humidity extremes, and high altitude.
The bottom line is that it might be best to consider your fingerstick blood sugar measurements as more of a ballpark figure than an exact one. It’s also a good reminder to trust your symptoms and err on the side of caution. If you feel hypoglycemia coming on, maybe don’t let a higher-than-expected blood sugar measurement prevent you from treating it.
Continuous Glucose Monitor Accuracy and Performance
The continuous glucose monitor (CGM) is considerably better than old-school blood sugar testing in so many ways: fewer finger pricks, constant updates, tons of actionable data, trend arrows, alarms, remote sharing. But do you have to trade accuracy for all that convenience?
Luckily, CGM accuracy is a somewhat less mysterious subject than glucose meter accuracy, because the manufacturers are obligated to run rigorous trials of their devices, and they are usually eager to publicize the results.
First, a few details. The CGM doesn’t actually sample your blood – it samples the interstitial fluid that surrounds the body’s cells. Interstitial fluid carries nutrients received from the blood capillaries, including glucose, which makes it a fairly reliable indication of blood sugar levels. But it doesn’t have exactly the same amount of glucose, which means that CGMs apply an algorithm that predicts blood sugar levels.
It also takes a little while for the glucose in your blood capillaries to filter into the interstitial fluid. When your CGM gives you a new glucose measurement, it’s really showing you what your blood glucose level was about 10 minutes ago. As a result, measurement disparities between two devices – or between your device and your body – will be especially large when blood sugar is changing rapidly.
Many CGM users also feel that their sensor is significantly less accurate during its first 24 hours or so, and the data backs this up. You can expect some accuracy issues that first day. The usual advice is not to aggressively calibrate it on this day, but just accept that the numbers might be less precise than usual, and allow it to warm up and improve like it’s been designed to do.
CGMs accuracy is judged mostly by MARD, which stands for mean absolute relative difference. To find a device’s MARD, researches take CGM readings and compare them to a reference, ideally a highly-precise laboratory blood sugar measurement. MARD is reported as a percentage: a MARD of 10 percent, for example, means that the CGM is on average within 10 percent of the reference value.
CGMs have gotten more and more accurate over the years. Earlier generations of CGMs commonly reported MARDs in the range of 15%, far less accurate than glucose meters are supposed to be. In those days, CGMs were only approved for “adjunctive” use – patients were not supposed to use the measurements to make insulin treatment decisions, and were advised to double-check with a glucose meter. Nowadays CGM manufacturers report MARDs under 10%, and CGMs readings have been approved for insulin management decisions, as well as for use in automated pump systems.
MARD has plenty of flaws. It doesn’t tell you what you want to know about a CGM’s accuracy in predicting trends, nor does it indicate how accurate a CGM is during period of rapid glycemic change (a study of the now-obsolete Dexcom G4 found that MARD roughly doubled during big blood sugar swings). But it’s what we’ve got.
CGM or Glucose Meter: Which is More Accurate?
Many people in the diabetes community believe that fingersticks ultimately give a more accurate reading. That’s understandable. For one thing, some CGM systems will actually prompt you to calibrate your CGM sensor by using a glucose meter, which seems to imply that the glucose meter is the more reliable device.
But as we learned above, the real-life accuracy of glucose meters leaves something to be desired. So which device is more reliable?
Given the extreme variability found among commercial glucose meters, it’s impossible to give a general answer to this question. But this 2018 analysis – which compared both CGM and glucose meter performance using a single consistent metric, MARD – offers a few insights:
The best glucose meters are still more accurate than any CGM. The Freestyle Libre 3 and the Dexcom G7, the two most hotly anticipated products of the next generation of CGM technology, report MARDs of 7.9 percent and 8.7 percent respectively. But that 2017 study found half a dozen meters were better than this even this standard, which was produced in ideal laboratory conditions. It is possible that the real-life performance of CGMs is somewhat worse than the reported MARD scores.
Lower-quality glucose meters are probably worse than any CGM. The same study identified ten glucose meters with MARDs above 13 percent, far worse than any CGM on the market today.
The CGM MARD numbers are arguably more trustworthy, because CGM accuracy studies are designed to satisfy an intense FDA approval process. The real-life performance of CGMs is also closely scrutinized by customers. The glucose meter market, by contrast, is sort of a wild west, with many manufacturers operating under relatively little government oversight. We know that there are no unsavory CGM manufacturers releasing shoddy merchandise, but we can’t necessarily say the same thing about glucose meters.
Both Dexcom and Abbott now claim to have created the world’s most accurate continuous glucose monitor. I don’t know if we’ll be able to say which device is the true accuracy champion.
The important thing is that both are well within federal accuracy standards, and even if the best glucose meter is better, both the Freestyle Libre and the Dexcom are accurate enough to mostly eliminate the need for fingersticks. As the authors of a 2021 study of CGM accuracy put it, “all current CGM devices are accurate enough for safe and effective diabetes management.”
Glucose measurement devices always include a significant margin of error. When your meter or CGM tells you that your blood sugar is X mg/dL, you should realize that it’s really a ballpark figure, and that your real level may be significantly above or below the reported number. The real glucose level is usually, but not always, within 15% of the measurement.
The best glucose meters (“fingersticks”) are probably still more accurate than CGMs, especially when blood sugar levels are rapidly changing. The worst glucose meters, however, can be shockingly inaccurate, and many on the market appear to fall short of international accuracy standards. Unfortunately, there’s no good current objective data on which blood sugar meters are best.
CGMs, which are held to higher regulatory standards, are more likely to be reliable than shoddy glucose meters and are considered accurate enough for insulin dosing decisions.