Six of the very best – digging additional into the statistics

On this a part of the “Six of the Greatest” protection, we have a look at the 15/15 information from this take a look at, and the share of readings the place the CGM didn’t present a low worth when finger sticks and hypo signs prompt in any other case.

The info on this piece is far more open to problem for 2 causes. The numbers of factors when there have been values out of vary have been comparatively small (11 under vary for Dexcom and 13 for the others), and never solely do CGMs change into much less correct when out of vary, so do blood testing gadgets, so there’s a higher margin for error on either side of the comparability.

15/15 take a look at

What’s the 15/15 take a look at? 

What you’re on the lookout for is the share of datapoints which might be inside 15mg/dl when glucose ranges are under 70mg/dl (3.9mmol/l) and 15% of the blood studying in any respect different instances.

This information signifies how correct CGM programs are when in a harassed state as a consequence of decrease or larger glucose ranges. On this instance, it gives a sign of how good they’re at these totally different ranges, however the variety of factors may be very low, so this also needs to be considered when reviewing this.

The desk above exhibits pretty dramatic variations between the totally different programs. The G6 and ONE have a lot larger percentages of their datapoints throughout the boundaries laid out when low or excessive than any of the opposite sensors. What’s maybe barely shocking is that in “in vary” durations, the ONE carried out far more intently to 2 of the newcomers than we’d anticipate. Total although, it highlights the variations within the efficiency of the totally different programs at decrease and better ranges. 

Proportion of readings >3.9/70 when blood studying is <3.9/70

As we will see from the earlier part, not one of the newcomers carry out notably nicely within the hypo vary. This part appears on the share of outcomes that weren’t displaying as hypo after I had hypo signs and glucose ranges have been registering under 3.9mmol/l (70mg/dl) on the fingerprick.

As I’ve already talked about, there’s a lot higher margin for error with these numbers, nevertheless, what they seem to indicate is that both some sensors weren’t pretty much as good at selecting up low glucose ranges, or that the lag on some sensors was a lot higher than on others. Whichever approach you have a look at it, the info within the desk speaks for itself.

Any conclusions from this?

As I’ve talked about, because of the higher threat of systemic error on this information, it’s not truthful to state conclusive outcomes, nevertheless, indicatively, these outcomes really feel about proper as a person, and again up the error grids and MARDf information. Additionally they have a tendency to spotlight the problems with a few of the accuracy research, the place nearly all of the contributors maybe didn’t expertise as extensively various glucose ranges as many individuals with T1D do.

Realistically, this provides to the dataset relating to the alternatives that folks make relating to which system they need. I’m personally not shocked by the end result, and suppose this backs up my desire of sensor. 



Leave a Reply