I will preface this blogpost by reminding you that I am not a medical professional and I am not suggesting that you do what I do.
Setting the Scene: On July 21 Dexcom will have an Advisory Panel hearing with the FDA to discuss whether Dexcom’s G5 CGM should be officially approved for dosing insulin. (This Diatribe article gives information on the meeting and how to provide comments or attend in person.) Currently the G5 has “adjunctive” labeling which means that it is approved by the FDA to be used with confirmatory finger stick testing. Non-adjunctive labeling will help CGM’s receive approval for Medicare coverage as well as justify CGM use for more people with diabetes. In September 2015 the G5 was approved with proper calibration for insulin dosing in Europe.
Background: Three years ago I wrote a blogpost about how frequently I dose off of my Dexcom. In that post I mentioned bolstering my Dexcom confidence with an average of 8 meter tests a day. Not much has changed since then except I now back up my Dexcom with only 4-6 daily finger sticks. Two of those BG tests are for calibration so it is evident that I am relying on Dexcom for most of my BG information.
Dexcom Blindman’s Bluff: Earlier in June I decided to test my confidence in the Dexcom G5 and play a game that could be called Dexcom Blindman’s Bluff. The purpose was to experiment with reducing the number of meter tests even further and to consider whether I have enough trust in a CGM to allow it to be the basis for all insulin dosing as it might be in an artificial pancreas (AP) system.
Articles about improvements in AP technology seemed to be published daily. Many clinical trials report average blood glucose numbers in the high 120’s to the low 140’s. For the vast majority of people with Type 1 diabetes, those numbers would be a significant improvement.
I am a T1 who has managed to have lower averages than that for many years. Would I be willing to trade a higher A1c for a huge decrease in the mental burden of diabetes along with a significant reduction in variability and the number of lows? I suspect that I would although I know some T1’s who definitely would not. Secondly, would I trust a Dexcom sensor to give sufficiently accurate results to dose insulin without my input?
For 48 hours on June 7-8, I decided to accept all numbers in my Dexcom target range of 70-160. The rule was No peeking at my G5 screen for any reason other than to act on prompts for calibration and alerts for out-of-range BG’s. I would use my Freestyle meter only for calibrations. I would rely on Dex alert numbers as the basis for correcting lows and highs and no meter tests for confirmation were allowed. Before eating I would not check my Dex for my current BG status and I would dose insulin based only on the carb count of the food I was eating and my anticipated activity level. If I hadn’t received an alert, I knew that my BG was somewhere between 70 and 160 and that was it.
Because I had Dexcom alerts and calibrations, I didn’t believe that I was doing anything horrendously dangerous. After all, those of us diagnosed with Type 1 diabetes before the mid-1980’s had no home blood glucose monitoring and played “chicken” with diabetes every day.
Results: I had no problems blindly following my Dexcom G5 for 2 days and I was comfortable letting Dexcom run the show. That confidence came with a couple of caveats. The sensor was a week old and had already proven itself to be reasonably accurate. I was close to home both days and not climbing mountains, playing golf, or eating chocolate-covered donuts.
I am the first to admit that my results were skewed for the experiment because I tried to avoid anything that would push my BG out of range. It was two days of the best diabetes numbers that I have seen in ages. But good numbers weren’t the purpose of the experiment and when I got a couple of low alerts overnight, I treated them with no meter confirmation and the same with an afternoon high.
This experiment reinforced how little I can feel differences in my blood sugar. Most of the time I had no clue where my BG was in my range. 75 or 155? I had no idea.
Although a two-day test is not sufficient to prove that I would trust the Dex as the compass for an artificial pancreas, it did show me how quickly I could adjust to accepting all numbers in a target range and letting a tech device deal with the minutiae. The idea of not thinking about blood sugars all of the time was very enticing and a slightly higher A1c, especially with less BG variability, might be a small price to pay for a more “normal” life.
Longterm: Two days was only a toe in the water of using a CGM as my primary blood glucose guide. I thought about doing a longer test, but I couldn’t think of any reason to continue using a CGM blinded. Trends, spot-checks, and pre-meal info are crucial features of CGM use and it is an artificial, somewhat silly game to do without them.
Even when personal experiments aren’t scientific and don’t prove a lot, they definitely relieve the tedium of diabetes and that is something I need. A more apt experiment for testing the G5 as a non-adjunctive device would be where I used my meter only for calibrations and as Dexcom recommends in Europe, when BG is low or changing rapidly. I don’t need to do that test because that’s close to what my life is right now.