For years now I have been avoiding CGM data hassles by overlapping the start of a new sensor with the end of the previous one. I first wrote about this back in 2020: Minimizing CGM gaps.
This does mean I essentially use two CGM programs, one tracking the new sensor until I decide it’s trustworthy and then switch my AID system over to using it. But I find it massively useful.
I often mention that for my body Dexcom sensors (both G6 and G7) the 30-minute, 50, or even 2-hour warmup outages are not useful. It typically take a whole day for a sensor to settle in, even if it starts producing what looks like reasonable numbers after a few hours. I wrote an article about this (Dexcom G6 warmup) back in 2022. I’m led to believe from discussions with users across the world and with other industry insiders that this sort of thing happens to a certain proportion of users. Some people are lucky enough to have bodies that the sensors settle in very quickly.
Having just started a new sensor this weekend, I thought it might be educational to illustrate again what actually happens. The following is not unusual for new CGM sensors for me.
First day issues
As you can see, the old sensor (blue line) was reasonably close to fingerpricks whenever I tested. These were done with Accu-Chek Guide and Guide Me meters.
But it took a whole 24 hours before the new sensor settled in over at the right of this graph.
It did settle in by itself: no calibrations were applied to it. So that’s nice.
But during that first night it spent a lot of time reporting hypos. Including gaps where it didn’t return any data. I disable all alarms from this sensor until it’s trustworthy, or my family would have got no sleep. Alarms are of course in place on the “live” sensor.
I know to not pay attention to the new sensor until it’s getting close to 24 hours old.
At the right side of this graph the old sensor was 10 days old. Luckily because I’m using an Anubis transmitter it will keep going past this, which gives me a bit of a safety buffer if the new sensor failed/etc. For me they are usually reliable out to around 14 days (sometimes more, sometimes less). By that time I need the new sensor to be reliable so I can switch to it.
Luckily where I am we’re subsidised for enough sensors to sneak a day off the official 10-day lifespan of G6 sensors, and start the new one slightly early. As a result I end up with CGM data that’s very reliable. If I was stuck trying to do something useful with the new sensor in that first day, life would be a lot harder!
What about G7?
As I mentioned: G7 sensors display similar startup behaviour for me.
Earlier this year I used a string of six G7 sensors in parallel with my G6s, but unfortunately found a string of issues with them. The locked-in cut-off at 10.5 days I could cope with, but other factors mean that I cannot rely on G7 sensors for useful AID data. The data jumps up and down a lot (“signal noise”) and the transmitter occasionally goes silent for up to hours at a time (before back-filling and pretending that the data was there all along). Not useful where the system needs to react and adjust insulin doses along the way.
But if the quality of the G7 sensors eventually gets improved to the level that I could rely on them after that first day, this same overlap method could deal with the extended “warmup” required. Even though the shorter 30-minute warmup just exposes the bad data sooner, I at least get to see when it’s looking trustworthy.