Hi. Wondering why, when I calibrate my Dexcom, it doesn’t show the number that I enter. For instance, my Dexcom reading was 113. My glucometer said 130. So after entering 130, the Dexcom said 121. Why doesn’t it say 130? Thanks.
The Dexcom CGM uses a mathematical formula, an algorithm, that considers not only the current blood sugar number but also recent trends and rates of change. I know, sometimes the CGM calculation is baffling. Also remember that ideally, the CGM reading will lag the finger-stick number by about 15 minutes. In your example, did the Dexcom 121 ramp up to 130 in 15 minutes or so?
3 Likes
Sometimes it seems downright perverse. For example I have a sensor in at present that seems to be nearing the end of its useful life. Last night around 3 am alarms were going off, and sensor reading was flatlining around 4.2. Fingerstick test said 6.7 so I put that in as a calibration to silence the alarms and get some sleep. Sensor reading now went to 4.5! More alarms so tested again and fingerstick now 6.6. Put in another calibration and sensor reading went to 5.2. This morning sensor reading was 3.8 and fingerstick 3.9. Figure that one out!
I don’t know the particular details of the Dexcom algorithm, but it is likely as @Terry4 points out that it is some sort of moving average or a filter related to a moving average. And we do know that the know that the moving average takes into account at least two calibrations since two readings are needed to calibrate a new sensors. My experience is that at least to a first order, the calibration results in closing the gap between the meter reading and the Dexcom by almost always half, which is what you observed.
And while you might be tempted to calibrate a second time to close the gap, that is generally a mistake. Your meter is only accurate +/- 20% and using the moving average actually averages out the errors and may enable the Dexcom to be more accurate than your meter. So hold off on the second calibration and trust the law of averages.
1 Like
@jjim335 - I get inaccurate night time low alarms too. When I relied solely on the CGM, I would just take a couple glucose pills, go back to sleep and the alarm would usually stop - but then my morning fasting would be very high. Then I started to double check with the glucometer and found the CGM to be wrong most of the time . So now when it starts acting up, and I don’t feel low, I turn the CGM off when it alarms. In general, I rely on the CGM primarily for trends. It usually reads low. I’ve set the high alarm at 130 in hopes of catching a high before it goes too far. Still wish there were a faster acting insulin.
1 Like
Thanks @Terry4 - yes, I’ve noticed that occasionally the CGM catches up to the finger stick. I have to remember the 15 minute lag.
Thanks @Brian_BSC !
Personally I almost always find it to be reliable overnight (some people suggest that sleeping on the sensor can cause inaccurate readings but I have never found this for me). The only time it is significantly out of whack is when the sensor is getting old.
It actually averages. Add 113+130= 243/2 = 121.5 or about 121. This is not a magical algorithm.
When my Dex is alerting about false hypos in the middle of the night, I’ll enter my reading in twice in a row. Then the Dex displays the number I entered immediately and is less likely to give me another false alarm an hour later. I like to think of it as the “I really mean it” mode. 
1 Like