We’ve all heard about the Libre Duo, offering glucose and ketones in a Libre 3+ sized sensor, but there hasn’t been much information on how it works or what the app looks like. I found a copy of the user manual, so here’s a bit of insight.
It’s the US app manual from September 2025, as submitted to the FCC during radio approvals, describing a sensor worn for up to 15 days. The FDA has since authorised a 10 day version, so the details are likely to have changed. I haven’t worn one. The screens are Abbott’s illustrations, and glucose is in mg/dL, so I’ve added mmol/l alongside.
What’s perhaps the most interesting isn’t really the app or what it looks like, but what the data from the studies shows. You’ll find that as you go through.
What does the app show?
Glucose takes up most of the home screen, with the number, arrow and graph you’d expect, and ketones sit in a strip underneath.
While ketones are normal that strip doesn’t tell you much. Below 0.6 mmol/l it’s green, says “Normal Ketones” and gives no number.

Once ketones reach 0.6 the strip changes colour and you get a number and an arrow. Here’s the manual’s example, at 17.8 mmol/l (320 mg/dL) with ketones of 1.1. Tapping or swiping up on the strip swaps the glucose graph for a ketone graph.

Here’s what you get at each level:
| Ketones (mmol/l) | Colour | Message | Number? |
|---|---|---|---|
| Below 0.6 | Green | Normal Ketones | No |
| 0.6 to 1.5 | Yellow | Elevated Ketones | Yes |
| 1.6 to 3.0 | Red | High Ketones | Yes |
| Above 3.0 | Red | High Ketones | No, it just says HI |
There are only three ketone arrows: rising or falling by more than 0.4 mmol/l an hour, or changing slowly.
The graph only covers 0.6 to 3.0, with anything lower drawn as a green line. That’s narrower than the 0 to 8 mmol/l we quoted back in June, although the sensor may well measure further.
Glucose works the same way as on a standard Libre, so I’ve left that alone.
What do the alarms do?
The glucose alarms are the same as on a standard Libre too. The app adds two ketone alarms at fixed levels, on by default, and to start with they sound even with the phone muted or on Do Not Disturb.

| Alarm | Goes off | Can you move it? | Can you turn it off? | Comes back? |
|---|---|---|---|---|
| High Ketones | Above 1.5 mmol/l | Manual shows no way to | Yes | Once per episode |
| Urgent High Ketones | Above 3.0 mmol/l | No | No | Every 2 hours until you’re back to 3.0 or below |
Nothing makes a noise between 0.6 and 1.5, where the panel just goes yellow. Above 1.5 you get the High Ketones alarm, and once it’s dismissed you won’t hear it again that episode. The urgent one above 3.0 repeats every 2 hours, and you can’t switch it off.


You can quieten everything with Silent Mode for up to 9 hours, including the urgent ketone alarm, although you may still get a vibration or notification depending on your phone.
How does that relate to blood testing?
Abbott says you won’t need a separate method to see your ketones, and nowhere does the manual tell you to check a ketone reading with a blood test. It doesn’t say what to do at any given ketone level either, beyond not ignoring symptoms and getting medical attention when appropriate. That plan has to come from your diabetes team.
The FDA’s wording is different. Its classification order says the system “replaces blood glucose testing for diabetes treatment decisions”. What it says for ketones is that ketone output is to be used “in conjunction with glucose levels, symptoms, and factors that may affect ketone levels”.
Abbott did compare the sensor with lab tests on venous blood, and overall the readings hold up well. In adults, 93.3% were within 0.3 mmol/l (below 1.0) or 30% (from 1.0) of the blood result, dropping to 88.6% on day 1.
The alarm figures are a different story as ketones climb. Abbott tested alarm levels of 1.0, 1.5, 2.0 and 3.0, although the app as described only has alarms at 1.5 and 3.0. This is how often an alarm at each level was matched by a blood ketone above that level, within 60 minutes either side:
| Alarm level (mmol/l) | Adults | Children 6 to 17 |
|---|---|---|
| 1.0 | 99.2% | 99.6% |
| 1.5 | 96.5% | 98.3% |
| 2.0 | 89.2% | 98.3% |
| 3.0 | 49.9% | 91.0% |
So the app’s first alarm is matched by the blood result nearly every time, and in adults the urgent one is matched less than half the time. It rarely missed when ketones were truly high, catching 94.9% of the 59 adult blood ketone events above 3.0, even though there were more alarms for numbers that didn’t reach 3.0. It looks like an alarm that errs towards going off early in Adults. That’s not dissimilar to the common view that the Libre Glucose monitoring tends to err on the lower side as well. What’s less obvious is why children had a much higher match.
The manual doesn’t say how far below 3.0 the blood values were when the adult alarm wasn’t matched, and someone at 2.8 is in a very different place from someone at 0.8. The FDA’s order sets one limit for adults: when the sensor reads above 3.0, no matching blood value can be below 0.6, which covers paired readings, not alarms.
It’s the distribution I’d most like to see. The classification order requires the labelling to describe ketone accuracy in four bands, including above 3.0, and this manual doesn’t, so the labelling on the shipped product will hopefully say more.
There are two points to be aware of on these numbers. At 3.0, the adult table counts 713 alarm events against those 59 blood events, and the manual doesn’t define an event, so these may not be 713 separate alarms that someone would hear.
It also doesn’t say which studies fed the table. Across five studies, ketones were raised by glucose manipulation, stopping insulin pumps, a low carb diet or a ketone ester drink, and the last two included people without diabetes. My guess is that most results above 3.0 came from the pump stops and the drink, but that is just a guess.
How long do they last?
The manual’s survival table is less comfortable than “up to 15 days” sounds. Sensors mostly get through the first week, and then they start to drop off.
| Still working at | Adults | Children |
|---|---|---|
| Day 7 | 94.2% | 89.8% |
| Day 10 | 87.0% | 78.3% |
| Day 15 | 71.1% | 48.3% |
Those figures leave out sensors lost for reasons Abbott considers unrelated to the device, such as being knocked off. Only 211 of the 356 adult sensors were still being counted on day 15.
Replace Sensor appeared early on 12.9% of adult sensors and 23.9% of children’s.
I don’t know why the product was authorised at 10 days and not 15, and it may simply be what Abbott asked for. I suspect survival played a part, because adult ketone accuracy held up (91.3% within 0.3 mmol/l or 30% on days 1 to 3, 95.6% on days 13 to 15, among sensors still working) while the number of working sensors didn’t.
For the 10 day product, Abbott’s announcement says 84.1% of adult sensors and 68.8% of children’s lasted the full 10 days, lower than this manual’s day 10 figures.
What does the 3.0 alarm mean in practice?
The app’s levels line up closely with NHS guidance on blood ketones, which describes its levels as a guide:
| Ketones (mmol/l) | NHS guidance for a blood reading | What Libre Duo does |
|---|---|---|
| Under 0.6 | Normal | Green, no alarm |
| 0.6 to 1.5 | Slightly high, test again in 2 hours. Call your team or 111 when you feel unwell | Yellow, no alarm |
| 1.6 to 3 | At risk of DKA. Speak to your diabetes team, even when you feel fine | Red, High Ketones alarm |
| Over 3 | May have DKA. Call 999 or go to A&E | HI, Urgent High Ketones alarm |
That makes the 1.5 alarm the prompt to talk to your team, and in my view it’s the most useful alarm on the app.
For the urgent alarm, I think you have to treat it as a blood reading above 3 and do what the NHS says, which is call 999 or go to A&E. This is one of those points where the statistics on alarm accuracy aren’t really in consideration. If you are above 3.0, then we know the alarm was generally right, and even if you’re not, it seems very unlikely that you’re going to not be extremely high, so presenting at A&E if you can, or calling 999 if you can’t, would be the sensible thing to do.
Some people will end up in A&E with ketones below 3 as a result. That’s not really an issue though. In my view, it is unlikely to be a problem if anyone between 2.5 and 3 is going to A&E anyway, even though the NHS guidance asks for a call to your diabetes team at that level. In children, 91.0% of these alarms were matched. In all these cases, caution is the right approach.
So what does this mean for you?
On a normal day it’s much like any other CGM. You get a green strip saying ketones are normal, and no number to think about.
The early stages are quiet. Between 0.6 and 1.5 there’s no alarm, only a yellow strip, and overnight you wouldn’t see it. If your clinic has recommended that you use the Libre Duo, then leaving the alrms on will be thr right approach.
Treat the urgent alarm as a blood reading above 3 and do what the NHS says. Silent Mode shuts the urgent alarm off, and all of the alarms need your phone on and nearby, so if you’re worried about ketones before or during sleep, this is something to be aware of.
Expect to lose some sensors early. On Abbott’s figures for the 10 day product, about one adult sensor in six and nearly one child’s in three don’t last, and a sensor that has stopped can’t give you a ketone alarm. It’s worth having a spare, especially when you’re unwell.
Overall, I’m not sure how much this changes things for most people. If you have issues with high ketone levels and DKA, or issues with pump failures driving those symptoms, having this combined sensor is likely to help. If you’re unwell, or worried about DKA, then it can provide a level of assurance. If you’re one of the many people that haven’t had DKA or registered particularly high ketones, it gives an extra layer of security that you’ll probably never use.
The choice, as Henry Kelly used to say, is yours!
Sources
- Abbott Diabetes Care. Libre Duo Continuous Dual Glucose Ketone Monitoring System: App User’s Manual. DOC53345 Rev. B, 09/25. FCC ID QXS-LIBDC1. Figures from the clinical studies overview and Tables 23, 24, 26, 27, 29 and 30; screens from PDF pages 19, 36, 40 and 41. All images are Abbott’s. mmol/l values are our conversions.
- NHS, Diabetic ketoacidosis, last reviewed 8 June 2023, accessed 31 October 2026.
- FDA announcement, 25 August 2026.
- FDA, De Novo classification order DEN250049, Libre Duo 10 Day, corrected letter dated 26 August 2026. The decision summary wasn’t on the FDA database when we checked on 31 October 2026.
- Abbott announcement, 25 August 2026, accessed 31 October 2026.
- Abbott, Libre Duo product page, accessed 31 October 2026.
- Diabettech, The Libre Duo: a platform prototype to move care from the hospital to the home?, June 2026.
Discover more from Diabettech - Diabetes and Technology
Subscribe to get the latest posts sent to your email.
Leave a Reply