To add more precision to the issue of dexcom accuracy:
- a glucose meter with strips has no memory, so every reading has the same likelihood of noise and bad data. And it gets worse when the BG is high or low.
- But since the Dexcom gets many readings to correlate to in its lifetime, even if it only lasts 7 days (2 days is enough btw), every time you give a strip reading (at least twice a day) it averages out the noise and refines its prediction. And, since each data point follows a previous data point, the algorithm can mathematically smooth the numbers and give you a more likely read.
So, in the end, in most cases after the first day (enough strip readings to refine and mathematically average out the mathematical noise inherent in the strip readings) the number on the dexcom is actually at least as good if not better than the strips. When I feel the calibration is good, I trust the Dexcom more than the strips. So does @rgcainmd whose opinion I value.
As for the dexcom being behind due to the interstitial liquid measurement vs the bloodstream: there is no doubt this is true, and I have seen it time and time again. There are two phenomena here you might be interested in as well:
even though it does take 15-20 minutes to be fully reflected, you will still see some influence from a change earlier than that. For instance, if you are steady and take some glucose, you will see an uptick on the dexcom within 5 minutes of taking glucose (but you would see a bigger uptick when measuring blood). So it is not that you don't see it for 20 minutes, it is that you don't see it all for 20 minutes.
When you have a sharp peak, up or down, you will see more extreme numbers on your strip than on the dexcom always. The Dexcom interstitial liquid averages out what it sees a lot, so a transitional peak will not be fully seen by the dexcom. Say your "real" blood readings in 30 minutes were, for instance, 135, 150, 175, 180, 165, 150 -> the Dexcom might see, eventually, a few minutes behind, 135, 145, 160, 160, 155, 150. So you may not see the full amplitude of sharp peaks. However, when you have a sharp peak, the read you get from the strip is highly suspect because strips are inaccurate high and low, so their reading would be suspect too.
So, on the whole, as I mentioned, in general I trust the Dexcom more, because, inherently, by the very mathematical nature of the operation that gives you the read, it should end up being more accurate - and in our case it is most of the time.
Sorry for the long post. As an engineer and scientist I tend to feel more strongly about the beauty of numbers and measures:-)