Sunday, 28 August 2016

Episode 2: Battle of the CGM systems; Part 1 The Analysis

Welcome to the battle of CGM systems in the Grand Canyon and Las Vegas. As a warning upfront, I have a lot to discuss so I have split this blog into two parts. You will need 15 minutes to read it, here a few pics for the warm up.

·         Part 1: This blog where I discuss the top line information and statistics with very little fluff, so the reader only interested in find out about the different CGM systems can consume without my anecdotes and stories.

·         Part 2: More detailed in terms of my opinion on the CGM devices, my exercise management strategies, download pictures of where different features of the CGM devices really came to the party, and my diabetes philosophy. It will also include a small scattering of stories from the trip including; getting engaged at the bottom of Grand Canyon, dancing up at the front of the Cirque Du Soliel Michael Jackson show, and lots of photos from the trip.

I have written this blog for a reader who has a good understanding of what type 1 diabetes is,  already knows what CGM is, has some understanding of the current products, and understands the different alarms you can use on CGM devices. If you currently do not have this understanding please link to the below articles, then this blog will be of much greater value:

For my reasons for wearing four CGM devices and the aims of the experiment please read the pre-trip article.  I was attached to four devices (Abbot Freestlye Libre, Dexcom G5, MinMed MM640G, Minimed VEO) for fourteen days whilst trekking rim to rim in the Grand Canyon and partying poolside at the Bellagio. Please remember whilst reading, this is a n=1 experiment and because the CGM devices worked in a certain way for me, does not mean they will work identically for you. Also some of the diabetes management strategies I use may be different to yours, so please question everything (I am my own diabetes expert, I am not yours) and consult your health care team before employing any of them. Disclaimers over, here are the questions I endeavoured to answer;

1. Which CGM system was the the most accurate for me compared to finger prick blood glucose (BG), and what is the practical impact of accuracy? Does every MARD % really make a difference?:
·         Which system has the most accurate MARD %
·         Which system has the most false alarms and missed threshold limits
·         What would be the consequence of using SG (sensor glucose) readings to correct at meal times in place of BG
2. What are the key features and unique selling points of each system and how do they trade off against each other?
3. What rank will each CGM system get from me in different key areas:

·         Accuracy
·         Ease of use (sensor insertion, accessing sensor information)
·         Key safety features
·         Key features for proactive management
·         Cost at manufacturers specification & at a practical level

4. My thoughts on the groups of people who may benefit the most from different CGM systems, and the NHS funding landscape

Let me begin.

1. Which CGM systems is the most accurate foe me compared to finger prick blood glucose (BG) and what is the practical impact of accuracy ?

      MARD (Mean Average Relative Difference) is the standard measurement of accuracy of SG readings compared to a blood glucose standard. The standard I used was the blood glucose readings from my calibrated and tested blood glucose machine (The Libre which uses Optium strips). Real life speak; MARD is how different on average the SG reading is compared to the BG reading. The first tale below aims to put this practically and shows the lower the MARD % the closer the SG range is to BG, which should translate into more effective diabetes management. The second table shows the MARD % of the four devices over the 14 days of wear, by showing accuracy in the first seven days of wear and if I could obtain readings after that.

Example SG range at different MARD% and BG levels
SG range
SG range
SG range
SG range
4.8 - 5.2
9.5 - 10.5
14.3 - 15.7
19.0 - 21.0
4.6 - 5.4
9.0 - 11.0
13.5 - 16.5
18.0 - 22.0
4.4 - 5.6
8.5 - 11.5
12.8 - 17.2
17.0 - 23.0
4.2 - 4.8
8.0 - 12.0
12.0 - 18.0
16.0 - 24.0

My 14 days trail MARD % for all four CGM devices (6th - 19th August 2016)

Device & number of sensors used in trial period
First 7 days after insertion
Days 7-10 after insertion
Advertised MARD % from manufacturers
Average (SD)
1 sensor
2 sensors
2 sensors
2 sensors
*Please note the MM640G I used had the old Guardian Link Transmitter, which has now been superseded by an updated version with a new algorithm that is reported to have a MARD of 9%. I was not aware of this until I returned home. Medtronic are sending me an updated Guardian Link Transmitter which I will wear and report back.

The table above shows the MARD % for the different devices. In the first seven days after sensor insertion, the manufacturer's suggested usage time for G5, VEO (6days), MM640G (6 days), it is clear that the G5 was the most accurate with a MARD of 10%, although all performed pretty much to manufacturers specifications. The Libre MARD was higher in the first seven days, but you can see the accuracy really improved after seven days. Unfortunately my perfuse sweating in the Grand Canyon heat of 40+ Celsius caused it to come off. I would not be surprised to see an overall MARD of around the 11% by day fourteen if the Libre would have stayed on. Both the MM640G and VEO use the Enlite sensor, and they both stopped working after 7 days and a second sensor was inserted for the next seven days. The G5 continued to work days 7-10 then died, I then inserted a second sensor. You can see that the accuracy of the G5 after day 7 deteriorated significantly, making the readings very unreliable.

My conclusion in terms MARD % accuracy are:

1. All CGM systems performed close to advertised accuracy and did a good job

2. The G5 was the most accurate in the seven days after insertion

3. The Libre showed great promise in accuracy for the full fourteen days, but keeping it in place was an issue, despite Skin Tac and Tegaderm. Please bear in mind I was in 40+ Celsius heat and when wearing in the U.K previously, the Libre sensors have lasted the full 14 days with a MARD of about 11-12%.

4. I could not get the sensors to go past advertised manufacturers time accurately. This may have been due to the environmental condition. I would not be keen to make therapy decisions based on the SG readings after seven days for G5 when MARD % increased, as you will see below, every MARD % counts when it comes to therapy decisions.

MARD %  means very little until you put it into real life situations, the main two being:

1. What impact does a MARD have on the number of false alarms and missed hypos?

2.  What happens if you bolus using the SG reading, do you suffers hypos and highs, does every MARD % matter?

I aim to run my blood glucose level 3.5 - 9.0 mmol/l to attain as near normal HbA1c as possible, usually ranges 5.8 - 6.4%, whilst preventing excessive hypoglycaemia. The download of BG readings below shows in the fourteen days I had four episodes <3.5mmol/l and only two readings >9.0mmol/l. The sensor downloads of the 14 days confirms the same picture.

Blood Glucose Download (6th - 19th August 2016)

CGM Download downloads for all four devices (6th - 19th August 2016)

I do not show these to gloat, but to show the power of having CGM to inform my diabetes decisions, and alert me when a glucose excursion is coming. This level of control was only possible because I set alarms on the CGM systems (you cannot set alarms on the Libre). I have evolved to using only the below alarms from years of trial error, they suit me and my management but may not necessarily be ideal for you:

·         Low limit was set at 4.2mmol/l on G5, MM640G and VEO. This meant every time the sensor level hit 4.2mmol/l on the different CGM systems, I got an alarm. This usually gives me time to take action to prevent a <3.5mmol/l, and does not annoy me with too many alarms in a day. I do not get huge swings and drops so i can get away without predictive alarms and not setting the lower limit to high. If i was a "swinger" i would have the predictive alarms on or set the lower limit around 5.0mmol/l. As you can imagine with three systems it was a good job they have a vibrate option otherwise my fiancé and other trek team members would not have been happy!

·         High limit set at 9.0mmol/l. I bolusing for food pre-meal by at least 15 minutes and choose meals that do not cause significant post meal spikes. If I was a Frostie muncher or sugar addict I would be setting this level at something like 14mmol.l. You need to make the alarms work for your lifestyle

·         As mentioned above I do not set any predictive alerts or rise/fall alerts because previous experience has taught me that this is overkill, and leads to extreme alarm fatigue! I have found from trial and error that if I set my low limit 1.2mmol/l above my hypo range I usually do not miss hypos.

·         I will discuss the use of Low Glucose Suspend and SmartGuard (predictive low glucose suspend) in the features section. They were not set routinely, only for demonstration purposes on certain days.

What impact does a MARD have on the number of false alarms and missed hypos?

I reviewed all of the downloaded information to look at how many false alerts at 4.2mmol/l and 9.0mmol/l I got with each system, and how many missed highs and hypos there were. The Libre does not have alert facilities, but i took the liberty of running the stats to show what would have happened if they did. The graph below shows the results:

Number of false alerts and missed highs and hypos for all four devices in 14 days
(6th - 19th August 2016)

*Please note the MM640G I used had the old Guardian Link Transmitter, which has now been superseded by an updated version with a new algorithm that is reported to have a MARD of 9%. I was not aware of this until I returned home. Medtronic are sending me an updated Guardian Link Transmitter which I will wear and report back.

Summary of the false alerts and missed highs and hypos:

·         Although  the Libre does not have any alarms I pulled the data to show what would have happened if it did. The Libre had a tendency to be running higher than the BG, evidenced by the highest number of false alerts at 9.0mmol.l e.g. the Libre was reading above 9.0mmol/l but the BG was below. There were only three false alerts at 4.2mmol/l, e.g. the Libre was reading 4.2mmol/l or below but the BG was higher. There would have been three missed hypos if my hypo level was 3.9mmol/l but none at my actual hypo level of 3.4mmol/l, so all hypos could have been prevented.

·         The G5 shows why a MARD of 10% vs 13+% is so important. The G5 had virtually no false or missed high alarms, only three false alarms at 4.2mmol/l and  no missed hypos at 3.4mmol/l. This shows that when it comes to alarms and preventing hypos, every MARD % counts.

·         The MM640G and VEO performed almost identically with very few issues with high alarms but a ten false low alarms at 4.2mmol/l, almost one per day, and the highest number of missed hypos at <3.5mmol.l, three and one respectively.

During my time as a medical device educator and now as a diabetes educator, I have started a lot of people on CGM systems. The number one annoyance reported back is the number of alarms and especially false alarms, this is why accuracy is key. If you get lots of false alarms, it is like the boy who cries wolf too often - when the real cry comes it is ignored and issues happen. The above data shows that every MARD % counts for preventing false alarms, and for me the G5 clearly comes out on top.

What happens if you bolus using the SG reading, does MARD % matter?

The G5 has both CE mark and FDA approval for people with diabetes to use the SG readings for treatment decision. The Libre has CE mark for similar as the G5 and has submitted to the FDA for approval. The VEO and MM640G do not have CE mark or FDA approval for the above at present. The wording of the approval for G5 and Libre is similar but practically it has very different application. I will show the detail from CE & FDA approval G5 and CE mark Libre, and explain in real life terms what it means:
·         *If your glucose alerts and readings do not match your symptoms or expectations, you should obtain a finger stick. A minimum of two finger sticks a day is required for calibration.
·         For children aged 2 and above and decisions to be made by a competent adult

Real life speak: As long as the user is confident in the SG accuracy when they do the two calibrations per day, they can without need for a confirmatory BG;
·         Bolus from the SG reading
·         Treat hypoglycaemia
·         Review data and change insulin doses accordingly
·         Make exercise plan adjustments based on the SG reading

·          (1) A finger prick test using a blood glucose meter is required during times of rapidly changing glucose levels when interstitial fluid glucose levels may not accurately reflect blood glucose levels or if hypoglycaemia or impending hypoglycaemia is reported by the System or when symptoms do not match the System readings 
·         For children aged 4 and above and decisions to be made by a competent adult

Real life speak: As long as the Libre does not have a an arrow going straight up or down the reading can be used for:
·         Bolus from the SG reading
·         Make exercise plan adjustments based on the SG reading
·         Review data and change insulin doses accordingly
·         If there is a arrow going straight up or down you cannot use the SG to calculate the correction dose at meal times, you must use a BG
·          If the reading says you are hypoglycaemic or hypoglycaemia is impending e.g. a SG reading of 5.5 with a straight arrow down, you must confirm with  BG

By just reading the internet you could be excused for thinking (I excused myself!) they both had the same clinical indications, but in practical terms they are very different. That might be due to the fact the G5 has a 2% better MARD as reported by the manufacturers. The data below will shed some light on what that meant to me, along with the already reviewed alarm data. I am not naive in thinking people with diabetes will follow the clinical indication to the letter, however, often there is a good reason why certain safety specifications are set. Let us see what it meant for me.

With this in mind I looked to see if using the SG reading from all the different devices at meal times to correct to a level of 5.6mmol/l would impact on my control. The measure I used was; for all of the bolus's in the fourteen days, what percentage of bolus's would end up in a specific target range the BG achieved:
·         within 1mmol/l of 5.6mmol/l (4.6 - 6.6mmol/l)
·         within 1.6mmol/l of 5.6mmol/l (4.0  - 7.2mmol/l)
·         within 2mmol/l of 5.6mmol/l (3.6 - 7.6mmol/l).

The below graph shows the results:
Percentage accuracy of correction bolus in target using SG of four CGM devices in 14 days
(6th - 19th August 2016)

*Please note the MM640G I used had the old Guardian Link Transmitter, which has now been superseded by an updated version with a new algorithm that is reported to have a MARD of 9%. I was not aware of this until I returned home. Medtronic are sending me a updated Guardian Link Transmitter which I will wear and report back.

Summary of using SG readings to bolus from:

·         If aiming to keep within 1mmol/l of 5.6mmol/l  (4.6-6.6mmol/l), using the SG reading would not be accurate enough for me. At best the G5 would hit that target 75% of the time and MM640G 60% of the time. This would mean one to two bolus's per day would go out of target.

·         If aiming to keep within 1.6mmol/l of 5.6mmol/l (4.0-7.2mmol/l), using the SG reading from Libre, MM640G, and VEO would not be good enough for me. At best I would hit that target 80% of the time, which would mean one bolus per day would go out of target. The G5 would be in the target 95% of the time and only one bolus in four days would go outside the range; this would be good enough for me.

·         If aiming to keep within 2.0 mmol/l of 5.6mmol/l (3.6-7.6mmol/l), using the SG reading from all SG devices would probably be good enough for me, as 90-95% would be in target, meaning one bolus every two to four days would go out of target.

·         This data again suggests that a small improvement in MARD % makes a difference in bolus precision for me. In practical terms for me;

o   I would be happy to correct to 5.6mmol/l  for G5. If I was a paediatric that would be 6.0mmol/l.

o   I would be happy to correct to 6.0mmol/l for Libre, VEO and MM640G and if I was a paediatric, 6.4mmol/l.

2.  What are the key features and unique selling points of each and how do they trade off against each other?

For each device I will discuss the key features that make the difference for me, along with the limitations and potential future alterations that would make them more suitable for me. I am not going to list every feature, you can go to the company's websites for that, and they are linked to in their respective sections. Each company has a variety of different data sharing APP's, receivers, and other innovations that I will touch on briefly. If you are into the data sharing on different devices, which is especially good for parents, please see the Nightscout website. In my opinion the mothers, fathers, grandparents and relatives that have made data sharing possible via Nightscout deserve praise of the highest order! I love #WeAreNotWaiting, tip of the cap to you all!

      What I loved:

1.      It was the easiest sensor to insert, it was the quickest to set up, and it was the quickest to start working at 60 minutes.

2.      No finger stick BG calibrations required due to their factory calibration method.

3.      It is also a blood glucose meter, ketone meter, and does have a bolus calculator if unlocked by a HCP code.  This has the potential to be a full all-in-one BUT, read the limitations part.

4.      The arrows showing the direction the glucose is going is very helpful in preventing highs and lows. Read part 2 on how I used the arrows to manage exercise, prevent hypos and change bolus amounts.

5.      As long as you scan every 8 hours you get a full profile of the full fourteen days. This allows you to:
a.      Use the Libre History Graphs to identify glucose excursion patterns that need therapy change, and to monitor progression.
b.      Download to the Libre Software or Diasend so a comprehensive analysis can be performed to optimise control.

6.      Now has an Andriod LibreLink APP so you can scan with your phone. Anything that goes on the phone is a bonus and I am never without mine, BUT, I am an Apple man so on the iPhone please!

The limitations and potential solutions:

1.      There are no alarms to warn if you hit a low or high threshold. A few times I scanned and I was already <3.5mmol/l and >9.0mmol/l. Whereas the CGM devices with alarms would have warned me at 4.2mmol/l and 9.0mmol/l, respectively. On the Grand Canyon trip especially, this would have caused me significant problems as I would have had to scan the Libre every 20 minutes to be safe, and it would have offered no protection through the night. To test this theory I turned all alarms off for two days to see if I could manage on the Libre scans alone, the download is below. You can see I missed a hypo on the 14th at 13:00 and a long one through the night on the 15th. For me, I manage my diabetes very proactively and the alarms at 4.2mmol/ and 9.0mmol/l are an absolute essential. However, for somebody else who wants to reduce the number of finger pricks, have trend arrows to help prevent issues, and not to be alarmed at during the day, this may be perfect. The obvious solution is for future models to have the option of alarms.

2.      It is almost an all-in-one as you can unlock a bolus calculator with a HCP code that allows you to put in carb ratios, sensitivity factors etc. This allows calculation of meal time bolus amounts. However, you cannot use the scanned Libre reading to give a correction and you must do a BG reading on the Libre to access it. Also, you cannot just enter carbs as you need to do a BG to activate the bolus calculator. This pretty much renders it useless for me as a bolus calculator. Obviously if you know your carb ratios, sensitivity factors etc, at different times of day, you can work it out in your head. But be careful here doing this with CGM because you will probably not be taking into account insulin on board when correcting in between meals, and the Libre does not show you this. I know a lot of people, including me, who have got giddy one hour post meal having seen a high SG reading and given a correction, later to be found chugging Lucozade! On a side note, in my opinion the ROCHE Bolus Advisor has the best bolus calculator as it takes into account both glucose centric and meal centric insulin and this allows a better calculation of true insulin on board. The obvious solution is to allow scanned Libre readings to be entered into the bolus calculator (as long as they are not very fast moving ones!), show insulin on board, and consider the algorithm used for calculating insulin on board.

3.      The fact it lasts 14 days is amazing but that is only amazing if it stays on for 14 days. Some tips I have picked up are:
a.      Put the sensor on the inside of the back arm so you do not bang or catch it on doors
b.      Use Skin Tac before applying the sensor
c.       Wear bandage, tegaderm or other apparel to keep it secure

4.      I love the fact you do not have to do calibrations, but if you are doing BG results anyway (confirm hypo, SG moving rapidly) it would be nice to have the option to calibrate, as this would help improve the accuracy. There is a Glimp APP that allows this, and anecdotally I have heard the MARD is better than 11%! See Nightscout for details. I am an Apple man so have not had the chance to give it a go but take a look here.

      What I loved:

1.      As discussed above, I loved the accuracy and it was my go to SG reading throughout the trip for making therapy decisions; preventing hypos and highs, planning carbohydrate intake for exercise, and giving corrections etc.

2.      When you calibrate you get an almost instantaneous shift in the SG reading to BG level, which gives the feel that it is back on track and increased confidence in the SG reading. This does come with some limitations as discussed below, but I did really like this, whether it is a Jedi mind trick or not!

3.      The receiver and display is small, the screen is easy to read and rate of glucose change arrows are easy to interpret. See part 2 for how I used the arrows to optimise control. My fiancé was most impressed with the leather pouch it came in that folds open, whilst I was more impressed with the accuracy!

4.      It has a G5 APP for both iPhone and Andriod that allows several devices to be linked to the sensor output. As I am an Apple man this means I can access the reading on my phone. Not really needed to put on my mum's phone but if I was 25 years younger this would be invaluable for a bedroom iPad or iPhone. See Nightscout for more details on further sharing capabilities.

5.      The G5 uploads to Diasend, which means I can add the data to that of other devices to get a complete picture of all the variables I need to assess.

The limitations and potential solutions:

1.      The G5 does not have predictive alarms, and I believe this explains why it can make an instantaneous shift in SG once calibrated. The G5 seems to prioritise preventing glucose excursions in the future by getting the SG back on track as quickly as possible, so the user can make the most effective future decisions. This is in contrast to the VEO and MM640G that use predictive algorithms which aim to prevent problems from happening, with predictive alarms and predictive suspends (SMART GUARD). Due to the predictive algorithms the change in SG once calibrated is much slower with the VEO and MM640G, but they do have the potential of preventing hypos more effectively before they happen. For me this is a question of user preference; are you a person that wants to be in charge and be very proactive with management? Or a person who would rather the technology make more of those decisions? As you may have guessed I am the former, and for that reason I prefer the G5 for that reason. It will be interesting to see in the future, with technologies moving to close the loop, if Dexcom incorporates predictive algorithms, or if it stays with the proactive user, or can it do both?

2.      When the G5 loses transmission with receiver for any length of time and then reconnects, there is no backfill of data. Therefore the transmitter is purely that, and it is not a recorder. This is in contrast to the Libre, VEO and MM640G which have about 8 hours worth of recording time. So as long as connection is not lost for more than 8 hours you get a full picture. This is very important when reviewing data on download, and could cause a serious issue for water sports people, and sports people who are out of range from their device for extended periods of time. Hopefully a future G5 transmitter will also have recording capability.

3.      At present the G5 is not compatible with Animas Vibe, only the G4. Therefore the G5 is not at present available in any integrated insulin pump system, and you would have to use a receiver or your phone. Carrying around multiple devices is a pain in the arse, and that is one of my favourite things about the VEO and MM640G, they are integrated and attached to me, so I always have the CGM to hand and arse! The quicker the G5 is integrated into the Animas Vibe, and potentially other pump systems, the better.

4.      The insertion of the sensor is the most tricky of all and is a bit "fiddly". A simpler insertion would certainly be welcome, but for me this is a minor issue when you consider the accuracy!

      I must mention that I worked for Medtronic for four years (2011 - 2015) as a Diabetes Clinical Specialist, and I did a series of blogs called 64 days on the MM640G about eighteen months ago, which you can look into if you like on earlier episodes of this blog. I was wearing the MM640G up until January 2016, but after a review of almost a year's worth of data I found that the sensors were only lasting 3 days on average, due to getting "Sensor Error" alerts, because of some issue with the Enlite sensor or new algorithm in the Guardian Link Transmitter. Therefore I changed back to my trusty VEO, which I love.  That being said it seems this "Sensor Error" issue may have been sorted with a change in the Enlite sensor, as I have been informed by Medtronic that a change in the Enlite was made recently, and during the 14 days I did not get any "Sensor Error" alerts.

Medtronic informed me after my trip that the Guardian Link Transmitter has been
updated with a new algorithm that boasts a MARD of 9%. They are sending me a new
transmitter with a new inserter to trial. Therefore the review in this blog and below is based
on the Guardian Link Transmitter that reports a 13-14% MARD and it may be that the new
version changes things. If it does what it says on the tin, I will  soon be back on the
MM640G. I will keep you posted because if it does do the trick then if you have the old
Guardian Link for MM640G, you will certainly want to be upgrading!

What I loved:

1.      It's a fully integrated system so you only need one device. It is connected to you and it is waterproof. Therefore there is no need to take it off and you have access to the CGM 24/7!

2.      It has 8 hours of data storage in the Guardian Link Transmitter, so even if the MM640G does lose connection you get backfill of data.

3.      It is easy to see in the night, so no need to turn on light or screw your eyes to see it. Very useful in a tent at the bottom of the Grand Canyon!

4.      SMARTGUARD during the night! When the only variable causing glucose drop is basal insulin (e.g. through the night), suspending the insulin 30 minutes before without alarms is perfect for preventing hypos and allowing a full night's sleep! This comes with a caveat - only if the SG reading is accurate enough. The accuracy section showed the MM640G had the highest number of false low alerts at 4.2mmol/l, which would have meant a lot of false suspends of insulin if I was wearing it for real. This is why accuracy is so important, and at a level of 14% MARD, it is questionable. I did not have SmartGuard working for me during the 14 days as I was wearing the VEO. In my previous Blog "64 days on MM640G" you can see lots of times where night time hypos were prevented. However, as mentioned in the alarms section with 10 false alerts at 4.2mmol/l, what woudl that have translated into false suspends. If the new Guardian Link Transmitter achieves 9% then Boom Shakalaker. I will report back when I have road tested it.

5.      With Nightscout you can share data just like the others; where there is a will, there is a way!

The limitations and potential solutions:

1.      SMARTGUARD during the day for me is not useful. Partly because of the current accuracy of 14% MARD, but mainly because only stopping basal insulin has no benefit when you add in bolus insulin, activity and food. In fact it is detrimental as it makes my insulin depleted later leading to rebound high blood glucose levels. See the download below where I turned SmartGuard on for one of the days hiking, to simulate what would have happend, remember I was always attached to the VEO for insulin. The blue markers indicate the basal would have been off virtually all day which would have lead to a lot of high glucose reading later. My solution for me is to only have SMARTGUARD on during the night and off during the day, which is great as the MM640G has that capability. I really hope the new MARD of 9% comes true for me with the new transmitter, as i believe SmartGuard is a massive safety enhancer but only at a MARD of 10% of 14%.

2.      You can only upload to CareLink Personal and as a patient you cannot get access to CareLink Pro reports. Diabetes is a SELF-MANAGEMENT LONG TERM CONDTION where the person with it needs to be empowered and educated to make all therapy changes. By providing the person with sub-standard reports this is in opposition to that. You can probably feel my passion on this subject, and this was the same passion I shouted from the roof tops when working for Medtronic. So again, please make the CareLink Pro reports available, or better still join Diasend so all of a person's devices can be uploaded to one place so that ALL PEOPLE WITH DIABETES HAVE THE BEST CHANCE OF SELF-MANAGEMENT. Rant over, I do think Medtronic is an excellent company and is one of the only companies who puts their devices to the test in Randomised Control Trials that allow re-imbursement to happen. Whilst other companies ride on the back of this evidence wave to bring their products to market, but that's capitalism for you!

3.      The time it takes for the SG to catch up with BG after calibration is frustrating. However, as discussed in the G5 section it is needed so that features such as SMARTGUARD are possible.

What I loved:
1.      Essentially, all the same things as the MM640G.

2.      I have a special place in my heart for the VEO as it has been literally by my side hugging my bum for the last 6 years! This is a testament to its quality.

3.      It has the option of Low Glucose Suspend that is very seldom needed for me, as the alarm at 4.2mmol/l alerts me so I can take action. However, on the day the two days I turned all alarms off to see If I could manage by Libre alone, I left Low Glucose Suspend on for back up. The download below is  the night after we had finished the Grand Canyon Trek, I was exhausted and fell asleep. That night I woke up to Low Glucose Suspend shouting at me. On review I had been hypo for ages and slept through 90 minutes of insulin suspension and Low Glucose Suspend alarm with a BG of 2.4mmol/l. Without Low Glucose Suspend that could have been much lower, and required my newly crowned fiancé Dani to use her Glucagon training! Secretly I think she would have enjoyed playing nurse. Low Glucose Suspend and the research evidence behind it is why the NICE (2016) DAP specifically indicates the VEO for funding for people with hypoglycameia unawareness issues.

The limitations and potential solutions:

1.      Same as MM640G for CareLink and accuracy.

2.      Although trusty, loyal and my best friend, it's not the most beautiful pump in the world and the screen can be tricky to see, especially at 4am! This has been rectified by the MM640G with the colour screen.

3. I will rank each CGM system on different key areas:

Accuracy for me: MARD, Alarms, safety to bolus from SG
1.      G5
2.      VEO
3.      Libre
4.      MM640G *pending trailing the new Guardian Link with reported 9% MARD

Ease of use for me: Set up, sensor insertion, navigation, looking at graphs, calibrating
1.      Libre
2.      All equal: MM640G, VEO, G5

Key Features for safety for me: Alarms, Low Glucose Suspend, SmartGuard,
1.      VEO - only one missed hypo and saviour of via Low Glucose Suspend on the 15th!
2.      MM640G and G5 joint second
a.      MM640G - SmartGuard during the night only for me but questionable with current 14% MARD. I  await to see if 9% with new Guardian Link makes a significant difference for me, then this would put it on top.
b.      G5 - no missed hypos with alarm set at 4.2mmol/l and amazing accuracy
3.      Libre - only get safety with extreme frequent swiping when doing heavy exercise and no protection through the night.

Key Features for proactive management for me: Making decisions in real-time for bolus,
exercise management, hypo prevention:
1.      G5 - Accuracy is king for pro-active management when combined with glucose trend arrows low and high alerts
2.      VEO, MM640G - Hard to differentiate from Libre on accuracy and glucose trend arrows, but having low and high alerts prevents obsessive compulsive swiping I needed to stay on top of things with the Libre
3.      Libre - Having no alarms was a limitation for me.


I originally wanted to say which I thought was the best value for money, but this question is completely dependent on the perspective of the user and their priorities. So in its place I have modelled some cost projection based on the current cost. These costing are subject to the companies changing prices and offering start up deals. I have not ventured into use of one sensor a month and other scenarios because the research, my personal experience, and my work experience all show clearly; for CGM to be of benefit and effective, it must be worn full-time. This is because how you manage your diabetes on CGM is very different to BG alone, and this takes practice, practice, practice. The cost models are for:

1.      Start up costs for receiver and transmitter

2.      Annual cost of full-time CGM according to manufacturers specification

3.      Annual cost of full-time CGM according to stretching the sensor usage out to 14 days each. My experience of them working to specified MARD% after day 7 for G5, VEO and MM640G is not fantastic, however, I am aware from the online community that some people report sensors last from 14 - 21 days. So I have included this.

Set up costs for all four CGM devices
Already have pump
Inc. charger
Already have pump
Inc. charger

Annual cost of CGM according to manufacturers specification
Each last 3 months
Excluding charger
Excluding charger

Annual cost of CGM according if can stretch sensor life to 14 days
Each sensor lasting 14 days
Each last 3 months
Excluding charger
Excluding charger

A quick summary would show there is very little difference for the G5, VEO, MM640G as the fully alarmed CGM systems, all at roughly around £3,000 per year. The Libre is only a third of that price at £1,159 per year but has the limitations of no alarms. I will discuss who may benefit from which more in the final section.

If you are interested in just dipping your toe into the CGM field it would seem sensible to try with the Libre as it is the cheapest start up, easiest to use, will not annoy you with alarms and lasts the longest. However, if you want to manage your diabetes more pro-actively or suffer from problematic hypoglycaemia issues, you may well choose an alarmed system instead. 

On a note to the companies! What would make this much more accessible would be bringing down the cost, which I know goes hand in hand with increased usage and therefore reduced manufacturing costs! But do not forget, I used to work for one of the big companies and know how much profit they make, even though Medtronic certainly does pump a lot of money back into research and development. Some initial concessions that would help:

·         Dexcom to reduce the cost of the transmitter. They only last three months and cost £200 each. If you have made a device with such a short battery life, that cost should be born by you, as the user/NHS bears the cost of each sensor. A reduction to £100 would be a generous saving of £400 per year, effectively two free months of CGM!

·         Medtronic could reduce the cost of the transmitter, which has 1 year warranty and cost of £350. If they only cost £150 then a saving of £200 would be a free month of CGM for the user/NHS who bears the cost of each sensor.

·         If fairness Abbott have done a good job with this and that is part of their marketing appeal.

4. My thoughts on the groups of people who may benefit the most from different systems.

The final section, Whoop Whoop!

IMPORTANT, what to come is my opinion only! If you fit into these categories it does not mean the NHS or your health care provider will fund it for you, and your Health Care Team may not even be interested. It may provide you with some information to have a discussion with your team and it may even persuade you to pay for it yourself. Again, question everything I write with a critical eye and make your own mind up. I have put the type of person I think would benefit the most for each device. After that I briefly discuss the NHS guidance for the U.K readers to help understand the funding jungle!  

Libre - Who I think would benefit most
1.      People interested in CGM but still not convinced and want to dip their toe to see if it will benefit them.

2.      People who want to significantly reduce the number of finger pricks but are not as interested in full-time proactive management of diabetes

3.      Parents who want to be able to see what happens during the night with a swipe, rather than waking their children

4.      People who just want diabetes to be a bit easier without information and alarm overload

5.      People who have budget constraints. Some parents receive DLA (Disability Living Allowance) money and this could be used to reduce the burden of diabetes. Do not start shouting at me here, I can feel the daggers in my back, it is only an option.

G5 - Who I think would benefit most
1.      People who want to pro-actively manage their diabetes with the most accurate information e.g. sports people, high flyers & achievers, engaged and educated parents, and the planners of this world.

2.      People with issues of hypoglycaemia that respond to alarms or their parents respond to alarms

3.      People who want to spend time learning about how to adapt new strategies according to CGM output and alarms

4.      Ideal for people on MDI who do not want a pump or an integrated pump system

5.      People who do not mind carrying two devices if they are on a pump. This is until the G5 becomes available in some pumps

MM640G - Who I think would benefit most
1.      People who want their device to take away some of the decisions and relieve some of the decision fatigue of diabetes

2.      People with issues of hypoglycaemia that do not respond to alarms or their parents do not respond to alarms

3.      People who are thinking of going onto a pump and want an all-in-one integrated system and obviously those already on the MM640G

4.      People who want to move closer to closing the loop. As advances in transmitter CGM algorithms will happen over the four years, you would have a pump that will access some of these

VEO - Who I think would benefit most
1.      People who want their device to take away some of the decisions and relieve some of the decision fatigue of diabetes

2.      People with issues of hypoglycaemia that do not respond to alarms or their parents do not respond to alarms

3.      People who are thinking of going onto a pump and do not have access to the MM640G and want an all in one integrated system, and obviously those already on the VEO

4.      People who love the tried and tested with solid research evidence proving Low Glucose Suspend

What NHS Guidance is there and how may it help?
There are a few key NICE guidance documents that will help in your quest to obtain funding from the NHS. This is a brief summary of my understanding but please go to INPUT for much more objective analysis and guidance documents/templates that will really help!

Continuous glucose monitoring
1.6.21 Do not offer real‑time continuous glucose monitoring routinely to adults with type 1 diabetes. [new 2015]
1.6.22 Consider real‑time continuous glucose monitoring for adults with type 1 diabetes who are willing to commit to using it at least 70% of the time and to calibrate it as needed, and who have any of the following despite optimised use of insulin therapy and conventional blood glucose monitoring:
·         More than 1 episode a year of severe hypoglycaemia with no obviously preventable precipitating cause.
·         Complete loss of awareness of hypoglycaemia.
·         Frequent (more than 2 episodes a week) asymptomatic hypoglycaemia that is causing problems with daily activities.
·         Extreme fear of hypoglycaemia.
·         Hyperglycaemia (HbA1c level of 75 mmol/mol [9%] or higher) that persists despite testing at least 10 times a day (see recommendations 1.6.11 and 1.6.12). Continue real‑time continuous glucose monitoring only if HbA1c can be sustained at or below 53 mmol/mol (7%) and/or there has been a fall in HbA1c of 27 mmol/mol (2.5%) or more. [new 2015]
1.6.23 For adults with type 1 diabetes who are having real‑time continuous glucose monitoring, use the principles of flexible insulin therapy with either a multiple daily injection insulin regimen or continuous subcutaneous insulin infusion (CSII or insulin pump) therapy. [new 2015]
1.6.24 Real-time continuous glucose monitoring should be provided by a centre with expertise in its use, as part of strategies to optimise a person's HbA1c levels and reduce the frequency of hypoglycaemic episodes.[new 2015]

My opinion: A lot of people meet these criteria and would really benefit from full-time CGM. However, this NICE guidance does not mandate the commissioners to pay for it, it is merely guidance. Their decision to fund will be based solely on the strength of the individual funding request submitted by the health care team looking after the person with type 1. Therefore if you believe you meet the criteria, need it, and deserve it; you owe it to yourself to make your health care team believe also! That is what I did! But please remember your health care team do not make the decision, they work with you to submit the request, they will be working with you if they believe!

1.2.62              Offer ongoing real-time continuous glucose monitoring with alarms to children and young people with type 1 diabetes who have:
·         frequent severe hypoglycaemia or
·         impaired awareness of hypoglycaemia associated with adverse consequences (for example, seizures or anxiety) or
·         inability to recognise, or communicate about, symptoms of hypoglycaemia (for example, because of cognitive or neurological disabilities). [new 2015]

1.2.63              Consider ongoing real-time continuous glucose monitoring for:
·         neonates, infants and pre-school children
·         children and young people who undertake high levels of physical activity (for example, sport at a regional, national or international level)
·         children and young people who have comorbidities (for example anorexia nervosa) or who are receiving treatments (for example corticosteroids) that can make blood glucose control difficult. [new 2015]

1.2.64              Consider intermittent (real-time or retrospective) continuous glucose
monitoring to help improve blood glucose control in children and young people who continue to have hyperglycaemia despite insulin adjustment and additional support. [new 2015]

My opinion: A lot of children and young people meet these criteria and would really benefit from full-time CGM. However, this NICE guidance does not mandate the commissioners to pay for it, it is merely guidance. Their decision to fund will be based solely on the strength of the individual funding request submitted by the health care team looking after the person with type 1. This is slightly different if you are on an integrated pump with sensor (see below). Therefore if you believe you meet the criteria, need it, and deserve it; you owe it to yourself to make your health care team believe also! But please remember your health care team (this is me) do not make the decision, they work with you to submit the request, they will be working with you if they believe!

3. NICE (2016) DG21 on integrated sensor augmented pumps- For Children and Young People only - those 18yrs or under

“MiniMed Paradigm Veo system is recommended as an option for managing blood glucose levels in people with type 1 diabetes only if they have disabling hypoglycaemia”

“The Vibe and G4 PLATINUM CGM system shows promise but there is currently
insufficient evidence to support its routine adoption in the NHS for managing
blood glucose levels in people with type 1 diabetes”

“MiniMed 640G system has not been assessed in the guidance, and the recommendations, therefore, do not relate to its routine use in the NHS”

My opinion: A lot of children and young people are already on a VEO that meet these criteria, and there should is a direct line of funding from NHS England. So if you believe, get asking. If you are on a MM640G or Animas Vibe you may be able to slip under the radar and hopefully future evidence will extend to these.

Children and young people with type 1 diabetes who have frequent severe hypoglycaemia are offered ongoing real‑time continuous glucose monitoring with alarms.

My opinion: This is a bit of a game changer for children 18 years or under. For the first time it is recognised that MDI patients should not have to go on a pump to get the benefit of alarmed CGM. If given a choice of losing the pump or CGM, I would lose the pump and stay on CGM with MDI. I believe the information and lifestyle management possible with CGM outweighs the advantage of a pump. Put the two together BOOM, but if a choice, CGM for me. There is no direct line of funding for this at present and each patient will have to go through an individual funding request, unless your centre has already got an arrangement in place for a pot of money. Either way you have to believe and then make your health care team believe you need and deserve it.

On a side note, Abbott have put a request into the NHS to get the Libre onto FP10 for prescription. If this gets passed it means your G.P. will be able to prescribe it which would be a huge game changer! Watch this space!

Final comments

If you have made it this far then I hope it has been a worthwhile investment of your time (just imagine how much time I invested!). I believe CGM is the next big step forward for the majority of people with diabetes, not just those who can afford it, or the small niche who are currently benefiting from the NHS or self-funding. As with anything in life there is no 'one size fits all', it is horses for courses, and I hope this blog has helped you select the best noble steed for you.

I am going to write a blog over the next month (these things take serious time and I now have a wedding to plan!) on how I manage my diabetes with CGM using my USA experience as the case study. It will include:

·         How I use the trend arrows to plan for exercise

·         How I change bolus amounts based on trend arrows

·         How I prevent night time hypos with CGM

·         How I managed a 4 day trek with diabetes exercise management strategies

·         My life and diabetes philosophy and how stoicism and a growth mindset is my key to controlling Diabetes

·         Details on my trip:
o   Getting engaged in the Grand Canyon
o   Getting a ring together in the famous Pawn Starts shop in Vegas
o   Dancing at the front of MJ Cirque Du Soliel
o   How amazing the Wildland Trekking Company is, especially our guide Ken Parker

·         And much more

If you have found it useful please share with those who would benefit. Feel free to comment but do not expect a reply instantly from me. You will be better emailing me at if you want to know something specific. If you think I made the results up and want the excel spreadsheets your more than welcome to them, just email me.




  1. When I calculated the MARD on my 640G running with an updated transmitter and extended the life out to 10 days, it came back at 10.9%.

    I'd be interested to see what you get over the "correct" 6 days.

  2. This comment has been removed by the author.

  3. Hi John,

    I'm surprised to hear about all the issues you have had with Enlite sensors and SmartGuard for 640G.

    For me it is working flawlessly and 640G with SmartGuard technology is a game changer for me and in my opinion Medtronic has outpaced all competitors in the pump/sensor market with this pump and sensor technology.

    I have a MARD of around 5-8%, and am constantly bolus'ing of the SG value for meals and corrections, without any concerns at all.
    I calibrate 3 times a day, have the sensors on my thighs (I know many others are using it on the upper arm (like the Libre) with great succes! Try it!).
    I know that many have issues using the Enlite sensors on the abdomen, myself included. The first 6 months on Veo and sensor was a big messy affair, and I was about to give up sensor totally, but luckily someone told me to try using the thighs, and I haven't looked back since :) (I have tried a few times with the sensor on the abdomen after switching to 640G, but it's still a mess, with lot's of bad SG readings compared to BG readings.

    You write that Medtronic has released a new Minilink transmitter? Is this an official new release, any link to documentation about this? Or is it just the Transmitter with the G printet on it?

    ps: Love your posts, do more of them, especially when you go neerdy on the data analysis :)

    Best wishes from Copenhagen

    Kenneth Broxgaard

    1. Hi Kenneth,

      That's awesome that it is so accurate for you! Medtronic informed me they changed the algorithm in the guardian link 'recently' and as mine is a year old I think the update has happened since then. It has just arrived today along wth the new serter . My plan is to wear the new Guardianlink with the mm640g and the dex G5 at the same time for a final head to head. along with using dr Adolfson's bolus changes according to rate of change arrows!

      Fun times ahead!

    2. Hi again,

      thanks for the swift reply :)

      That's super interesting that they have changed the algorithm in the Guardian Link recently (mine is from fall 2015).

      What is Dr Adolfson's bolus changes?

      Looking forward to see the new test with the updated transmitter.

      Cheers !


    3. Update:

      I know a person who received a new transmitter just 1-2 weeks ago, and the version of her transmitter is 3.0A, where as mine, which is around 1 year old, is version 2.0C.

      So it seems like you could be more than correct in assuming that something big really has happened with the transmitter, when they raise the version number by 1.


    4. Indeed you are correct. A new version as of August 2016 with a new serter! Just started it and it is going much better than previous! I also have the reseat too so detailed blog to come!

      You may need to contact Medtronic for an upgrade! 😏😏😏