Freestyle Libre Accuracy vs SD Codefree

johnb46

Active Member
Messages
25
Hi, I have my first Freestyle sensor and its readings via the Freestyle LibreLink do not agree with my SD Codefree. The latter is 6 years old, has a new battery, has been recalibrated using the supplied dummy strip and gives readings that I have always believed in (!) and that have slowly worsened over the years. Moreover, during the 6 week period, approx, over which I have reduced my carbs to sub-30g/day from maybe 200g/day, it has looked convincing in the way that the 'spikes' from groups of vegetables have decreased as I decreased their amounts (e.g. my 3 egg veggie omelette's decreasing 70g mushrooms, 50g red pepper, 50g onion and 50g garden peas to 70/25/25 and 0g peas in stages). Maybe it's always been incorrect?
The Freestyle sensor, fixed to the back of my right arm like in the pictures - its top is half way between shoulder and elbow, tends to show a comparable reading to the Codefree before the above breakfast and shows a similar trajectory curve (reaches its peak at the same time, and returns to normal somewhat quicker, as the Codefree) but the maximum value is much lower, e.g. 8.1 vs 9.1 at breakfast and 7.5 vs 8.4 at lunch. So the peak rise at breakfast according to Codefree is +2.3, but only +1.3 for the Freestyle. A big %age difference.
I take the readings, 2 every 15 minutes, in the order Codefree, Freestyle, Codefree, Freestyle. The Freestyle ones are always the same and very quickly done, and the Codefree ones are almost always within 0.3 of each other; if not within 0.3, repeat.
Looking at each time point's readings, there is no simple multiplier that could be applied to the Freestyle that would make it match the Codefree.
One of the reasons for trying CGM is to find out whether my body is exhibiting aberrant behaviour when I'm not looking. But, of course, if its readings are inaccurate and muted it can still register unusual behaviour as if it were within normal range.
TIA. Any observations welcome!
(T2D, DX 2014, diet-controlled)
 

urbanracer

Expert
Retired Moderator
Messages
5,186
Type of diabetes
Type 1
Treatment type
Insulin
Dislikes
Not being able to eat as many chocolate digestives as I used to.
.......... So the peak rise at breakfast according to Codefree is +2.3, but only +1.3 for the Freestyle. A big %age difference.....................

Is it really though? Blood testing equipment that is sold in the UK has to conform to an ISO standard which allows a tolerance range of +/- 15%.

So let's say that your 'true' glucose level was 8.5 at breakfast time, the allowable error range would return readings between 7.2 and 9.7 and they would remain within specification.

Of course I can't know what your 'true' glucose value is but your meters are almost within 10% of each other so I'd say that's pretty good when all is considered. Remember also that the Libre measures glucose values in interstitial fluids and not actually in your blood. It tries to do a clever correlation but it's difficult and varies with every user.

We usually suggest to people that they should not get too hung up on individual numbers, just look for trends and see where you can minimise the peaks. The libre is great for this.
 
  • Like
Reactions: johnb46

type two paul

Well-Known Member
Messages
121
Type of diabetes
Type 2
Treatment type
Tablets (oral)
Dislikes
Diabetes
Is it really though? Blood testing equipment that is sold in the UK has to conform to an ISO standard which allows a tolerance range of +/- 15%.

So let's say that your 'true' glucose level was 8.5 at breakfast time, the allowable error range would return readings between 7.2 and 9.7 and they would remain within specification.

Of course I can't know what your 'true' glucose value is but your meters are almost within 10% of each other so I'd say that's pretty good when all is considered. Remember also that the Libre measures glucose values in interstitial fluids and not actually in your blood. It tries to do a clever correlation but it's difficult and varies with every user.

We usually suggest to people that they should not get too hung up on individual numbers, just look for trends and see where you can minimise the peaks. The libre is great for this.

I guess the HbA1c is the real deal.
 

johnb46

Active Member
Messages
25
I guess the HbA1c is the real deal.

Yes, as you say, the HbA1c is the real deal and, unfortunately, that is in line with my SD Codefree and way off the Freestyle's predictions (42 vs 33-35). Given there was little variation between the readings I got from Freestyle CGMs and that I don't get hypos and / or unpredictable deviations from my expected BG levels, I guess the Freestyle CGM is not for me.
 

Beansprout

Member
Messages
21
Type of diabetes
Treatment type
Tablets (oral)
I'm on my second libre. So, you know, tons of experience. ;) However, I will say the two I've used have been consistently below finger pricks. Within tolerance, interstitial.. probably.. still annoying as hell.

The first sensor was between .5 and 1 mmol lower on readings (based on the handful of checks I did throughout the two week period). The second has been closer to 2 mmol lower and thinks I'm in the 3s often (sooo unlikely and countered by finger pricks).

I was trying to correct the numbers manually at first (download xls file, adjust, upload to another app) but have given up. At the end of the day, the sensor is still really useful to me in tracking trends. I absolutely love having a continuous graph and have noted several useful things.

So, I've accepted it. I'm about to start a third sensor so we'll see what that one is like.

Bigger problem... There's no funding where I live and these babies are expensive!
 
Last edited:

johnb46

Active Member
Messages
25
I'm on my second libre. So, you know, tons of experience. ;) However, I will say the two I've used have been consistently below finger pricks. Within tolerance, interstitial.. probably.. still annoying as hell.

The first sensor was between .5 and 1 mmol lower on readings (based on the handful of checks I did throughout the two week period). The second has been closer to 2 mmol lower and thinks I'm in the 3s often (sooo unlikely and countered by finger pricks).

I was trying to correct the numbers manually at first (download xls file, adjust, upload to another app) but have given up. At the end of the day, the sensor is still really useful to me in tracking trends. I absolutely love having a continuous graph and have noted several useful things.

So, I've accepted it. I'm about to start a third sensor so we'll see what that one is like.

Bigger problem... There's no funding where I live and these babies are expensive!
Yes, mine showed me that I am a bit more sensitive to carbs at breakfast and I rearranged my diet. The Codefree would have shown me the same thing, but I had almost always not had the patience and discomfort tolerance to make measurements throughout the day. Diff between Codefree and Libres was a full +1mmol (e.g. bpost-breakfast peaks of 9.3 and 8.3, respectively). It's made breakfast less interesting than before!
 
  • Like
Reactions: Beansprout

KK123

Well-Known Member
Messages
3,967
Type of diabetes
Type 1
Treatment type
Insulin
We usually suggest to people that they should not get too hung up on individual numbers, just look for trends and see where you can minimise the peaks. The libre is great for this.

Hi, I do agree but it's hard not to get too hung up on individual numbers and it's those numbers that are guiding us to what amount of insulin to take for a meal. Once you are on the libre you get very few test strips and when I queried this they said you are to use the libre for all doses. My readings tend to be in the 4s and 5s so I think the libre is a bit out. I am waiting for my first hb1ac since the libre so hopefully I'll be able to gauge the accuracy.
 
  • Like
Reactions: urbanracer

michita

Well-Known Member
Messages
479
Type of diabetes
Type 1
Treatment type
Insulin
I also use SD code free meter as my 2nd back up meter. Ive tried libre (maybe 2 sensors a year). I actually quite like SD code free meter. Between SD codefree and libre, I would trust SD code free readings. Libre is good for seeing the up and down trends but I thought it’s often reading too low than actual,