A lesson in how not to roll out real-time CGM to an existing system

A lesson in how not to roll out real-time CGM to an existing system
A lesson in how not to roll out real-time CGM to an existing system

Approximately a month ago, Aboott started to make noises about the roll out of an update to the Libre2 software, following various approvals of their product for use as a Real Time CGM (mostly for competition in low cost markets with Dexcom’s ONE).

Essentially, the new update would officially allow both iPhone users and Android users to receive CGM via the LibreLink app (as many third party apps have offered).

Information was shared via senior UK diabetes charity and medical representativea as well as to users and advocates through the community that Abbott has formulated (And hands up, I am part of that community and welcome information through it). Essentially the news was that the roll out would start on the 12th July and would take about 7 days.

And that’s where it all went a bit Pete Tong.

Apple devices

The software roll out to apple phones was flawed. It didn’t work properly.

It deleted all history on your phone, when it worked, and mostly left users with a white screen and no information.

They lost local data, suffered a lot of pain and were forced to go back and try again, from their installed apps menu.

After a number of days, Abbott released a upgrade that solved the problems, but most shouldn’t have been there in the first place. 2.10 to is quite a lot of fixes…

More amusingly, perhaps, Medcity managed to write an article that made a positive out of the support provided over social media that should never have been required.

And what of Android?

The inconsistency of the Android roll out

Apparently, at the same time, the same software was being rolled out to Android users.

Though you could have fooled many of us, as we eagerly awaited rtCGM through the official app. At the time of writing my app remains on its previous version.

We were told that this would be phased, and yet phased seems to be code for “it appears that not everyone will receive the update”. A poll run on the UK Freestyle Libre users Facebook group suggests that more Android users have not received or been offered an update than have received it. In the poll, nearly 61% of users said they’d not seen it. Granted, an internet poll is easily biased, but that number doesn’t look great.

Abbott have not been clear on what they were doing, and leaving a significant number of users without the update also raises questions about what has been released.

Communication from them included the following:

  • The Android app roll-out has been phased over a week. Version 2.10 should be in the Google Play Store for all Android users soon (16th July)
  • Expect to see [the update] soon (19th July)
  • Abbott recently piloted version 2.10 of the FreeStyle LibreLink app for Android (23rd July)
  • The next version of the FreeStyle LibreLink app for Android will soon be available for download on the Google Play Store soon. People who have successfully upgraded to 2.10 or use a previous version can continue to use their app, and new users can download available version 2.8.4. (23rd July)

Which looks suspiciously as though there had been a change of plan with regard to the roll out of rtCGM.

And then on 26th July, a notification of a pause in the roll out due to issues reported with the software appeared.

They might be trying to avoid a further fiasco, but we may be beyond that point!

Where does that leave us?

It’s a difficult question to answer.

From what we can see in this case, you either have an updated app to allow rtCGM or you missed the cut. And if you have one, it probably didn’t work in the first instance on either type of phone.

If you missed the cut, it’s not clear when you will get an update (soon, we’re led to believe).

All in all, it has turned out to be a bit of a mess. And the people who have to deal with this are those who were supposed to benefit.

It all supports the view that this rollout was very much a pilot process, and had that been communicated to end users, we may have been a little more understanding. Instead, Abbott have launched a key new feature very badly, in a way that may still encourage users to desert for Dexcom ONE or other sensors.

Either way, many will continue to use third party apps to obtain rtCGM data off the Libre2, while Abbott sorts themselves out.

What could have been done differently?

Let’s not forget that Abbott already have a real-time CGM in the Libre3. One that we know works well. There was no need for them to convert the Libre2 into one.

Unless, commercially, they felt they had no choice.

With growth in Dexcom ONE in some markets eating into market share, has it become imperative to offer an equivalent?

Looking specifically at the UK, the cost level that the the NHS pays for the Libre2 is, we assume, lower than the Libre3. We know the price charged out to ICBs for Libre2. To roll Libre3 out to a larger number of users would entail bringing the pricing in line with the Libre2 system. Which, we assume, affects the bottom line for Abbott.

If we look at the UK data from OpenPrescribing.net up to May, we see that Libre2 issuance seems to have plateaued at around 190,000 per month. In comparison, Dexcom ONE has had a much lower uptake, but it is increasing.

In May, Open Prescribing shows only around 6,600 Dexcom ONE sensors being issued. Even with a projected growth rate on both of those, six months out the Dexcom ONE uptake would be expected to be significantly lower than Libre2, which would still be expected to be between 170,000 and 190,000 per month. In fact, at current growth, Dexcom would appear not to be touching Libre2 numbers for a number of years.

The graph below shows uptake of Dexcom ONE and Libre2 going back to the start of 2022 in the UK, normalising the Dexcom data to take into account that on a monthly basis, Dexcom requires three sensors to every two of Libre2. It also includes trendlines forecast out 6 months.

There’s a very clear difference in numbers, and one that, even at the current Dexcom ONE growth rates, doesn’t seem likely to unseat Abbott just yet.

More widely, replacing Libre2 with Libre3 to provide rtCGM would, perhaps, have forced the introduction of Libre3 in markets where it hasn’t been launched and there would be a variety of costs associated with that, so it was better to go to “enhanced” Libre2.

The question is, have the issues experienced in the UK cost Abbott more in both reputation and user take up, and ultimately, bottom line, than flipping to the Libre3 would have?

If we continue with the hypothesis that the UK was a pilot, then the pilot has done its job. It’s highlighted issues and put a plan in place to resolve them, opening up a much more straightforward roll out for the remaining territories.

From an end user perspective it’s not been a good look, and there are still a large number of disappointed people out there who haven’t yet got rtCGM. When they finally do receive it, will they forget about the trouble?

From Abbot’s perspective, reputationally, perhaps, they may have paid a price. For safeguarding future rollouts? Perhaps it was worth it.


  1. For those that did get the update, how reliable was/is the rtcgm data flow? A while back but I found xdrip/oop2 to be unreliable (lost data, losing connection). So I’ve still stuck with Libre2/miaomiao/xdrip/aaps-smoothing which is solid. Better than DexOne which I find uncomfortable and annoying to have to change mid-week.

    • I’ve only ever used Libre2 with alternative apps, so can’t comment on reliability with the official one. With the unofficial ones I’ve not had many issues.

      I’m absolutely sure that it’s generally the phone, rather than the sensors themselves that drive the issues though.

  2. Hi
    My partner is an ex Libre user now on Dexcom with a pump. Whilst I am technically competent, my partner sometimes struggles and I have witnessed first hand the rise of technology in Diabetes. I am incredibly grateful to the people who work on open software enabling my partner to manage her control. The thing that I really do not understand is why a company like Libre can push an update out with out a roll back process. Surely what should have happened is that as soon as it was realised there were issues, end users could easily select an option to revert to the previous version. Thanks

  3. Thought provoking article, thank you.

    If you’re right, apart from shooting the Dexcom One fox, and making it more difficult for a fraction of the 95,000 NHS patients that use Libre 2/LibreLink to keep their diabetes clinicians happy by proving data to Libreview whilst simultaneously using third party apps to have user interface friendly graphs, alarms, etc, what was the point of this exercise?

    Was it worth it? The data you’ve provided indicates that Dexcom’s One rtcgm is still making only slow progress into Abbott’s current UK market share.

    If your “guinea pig” hypothesis is correct and one is an insulin pump manufacturer looking to partner with a reliable cgm partner to take a share of the emerging and lucrative hybrid closed loop market, are you going to take Abbott seriously as a partner if it is prepared to conduct “guinea pig” trials of update software that didnt appear to have undergone any form of basic (alpha/beta) prior testing to ensure that it was relatively stable and fit for purpose, such as took place over the last two weeks?

    Was it worth the aggravation having to self-report to the MHRA the issues that have occurred for those “guinea pigs” (un)lucky enough to receive the update?

    Is this how testing of updated software for medical devices to be conducted in future in the UK? Would this hypothesied “guinea pig” update test be conducted for an update to a “life or death” software update (e.g. heart pacemaker app)?

  4. Although it seemed shambolic and distressing to those involved, I can report no issues with my iOS upgrade.

    However, after a few days use, I decided to stop bothering (other than to keep my DSN happy), switched off the alarms and reverted back to my Bubble/GlucoseDirect combo. Main issues: FREQUENT loss of connection and (long-standing) inability to calibrate and therefore I couldn’t trust the overnight hypo alarm would ring, or would ring inappropriately. Moreover, I prefer using the GlucoseDirect app, including the widget on my Lock Screen).

  5. Excellent article – interesting to see that Dex One is slowly taking hold – its still ‘challenging’ to get from Some GPs and ICBs, so when its truly a choice it’ll be interesting to see how threatened Abbots L2 foothold is. Even more so now it’s ‘CGM for all’ – so the uptake of Dex One may slowly win the battle (for now at least).
    Not the first time large companies have taken shortcuts to get a software version out so not surprised by that – more disappointed so many people lost the app and had to fail back to blood testing / another libre etc – always worth having a back out plan in case things go wrong.
    Looks like Abbott haven’t scaled production and/or got approval of the L3 yet for all markets yet – so the L2 ‘patch’ is to hold ground till its ready.
    Always thought it odd that if we can already do the ‘1 min update of BG’ to our phones with the Patched Libre app, or xDrip – why have Abbott taken so long to implement the changes – I guess they thought they had enough of the market not to have to do anything till the L3 was launched, so why bother – the majority of consumers will never know what they’re missing.
    Thankfully we have the Open Source community to push us further, faster – my life has been turned around in the last 5 years or so by what can be achieved, not what a big company is ready to sell us.
    Fantastic website, research and articles on here – thank you Tim (you really have changed my life for the better)

  6. A very good article. Thankyou. What frustrates me immensely is the lack of a straightforward response from abbott. In my eyes this is a complete mess up caused by insufficient regression testing pre roll out. I dont really want to entertain uncertified 3rd party solutions like juggluco, but you have to ask how release 2.10 ever got certified with it being unfunctional?

Leave a Reply

Your email address will not be published.