The Libre 3 and 3+ sensors proved wildly inaccurate and failure-prone for me. They often didn’t make it the 15 days of the sensor, constantly set off low-glucose alarms, were almost never close to accurate (100% difference, regularly), and Abbott was little help, when called (always talked to a Philippine customer service agent). They always blamed the phone, said I needed to go back to an earlier operating system. Ironically, I also used their reader, which did no better.
We were assigned the Libre sensor and reader by a veterenarian, for two pets; it ended up killing both of them. Wildly inaccurate information, for which we were to depend on providing insulin. I suppose we killed them…when we based our action on the data received.
I had both Dexcom G7 and the Libre 3+ in use at the same time, while transitioning to the G7; one in each arm, and compared them throughout the day and night with a finger stick glucometer reading. The G7 was more consistent, but only because it accepts “recalibration” with the fingerstick glucometer reading, and the Libre does not. The best I could do with the Libre is enter the fingerstick reading in a comment, which I always did…didn’t help with the accuracy, though.
I never viewed the Libre 3 (or the Dexcom G7) as truthful, accurate data; I view them as trend information that suggests how I’m responding to food, etc. I can make decisions with them, but I do so based on what I’m seeing on the sensor, AND what I see on a fingerstick glucometer reading, and when in doubt or with high or low indications, I’ll take several fingersticks, sometimes with several meters of differing brands, for verification.
The CGM takes data differently than a blood glucometer, so the readings will be different, and the CGM will lag fingerstick readings by roughly fifteen minutes. Fingerstick glucometer readings will vary; use the same reader in the same finger, you’ll get different readings. Different fingers, different readings. Different meters, different readings; one is always viewing a percentage of error, and shouldn’t consider any reading a hard value, but an approximation.