You may have seen me refer in the past to some of the clinical trials I’ve been involved in. Now it’s time for the next phase of news!
The “Baker AAPS” trial
In 2019 I started working with the Baker Heart & Diabetes Institute on a clinical trial of the safety and efficacy of the AndroidAPS AID system. It was a randomised crossover trial, although a small one with only 20 participants, all in Melbourne. The pandemic unfortunately introduced some delays, but the trial was eventually started and then in turn eventually concluded. In the end it’s taken some time for the clinical results to be published.
They were initially presented at the 2022 Australasian Diabetes Congress, which I wrote about back in August. But now the results have been published in the Journal of Diabetes Science and Technology. It’s good to get that milestone finally out of the way! I must admit I’m proud to have my name on it too.
Unfortunately the whole paper isn’t Open Access, but as one of the authors I’m happy to share details directly.
But even just the summary information as shown here tells most of the story!
The trial wasn’t studying the advanced features of the oref1 algorithm I’ve been using myself for the last few years but it’s a start. It was using oref1 just without bells and whistles like SMBs (micro boluses) turned on. It adds to the growing evidence around the world that these systems can be safe AND effective. The “Conclusions” spells it out:
This clinical trial of the open-source AAPS hybrid closed-loop system performed in an at-home setting demonstrated comparable safety to stand-alone pump therapy.
The glycemic outcomes of AAPS were superior with improved TIR, and there was no significant difference in TBR compared with stand-alone pump therapy.
“TIR” was Time In Range for 3.9-10.0 mmol/L. “TBR” is Time Below Range (not “Temporary Basal Rate”: there are only so many TLAs to go around).
The change in TIR from baseline for AAPS compared with stand-alone pump therapy was 18.6% (11.4-25.9), (P < .001)
That’s a difference between a TIR of 58% for people not using AID, versus almost 77% when they did. There was of course some variation amongst individuals, but that 18.6% is regarded as “statistically significant”. This was a proper randomised crossover trial, where everyone spent time in both modes.
And as always, this was nice to see:
No serious adverse events or episodes of severe hypoglycemia were recorded.
Keep in mind. many people around the world have built equivalent systems for themselves using AndroidAPS.
In 2022 researchers from the University of Otago (in NZ) published the results from their CREATE trial “Open-Source Automated Insulin Delivery in Type 1 Diabetes” in the NEJM. This was a larger trial using similar AID software (they did transition to using SMB/etc features). The pumps used in the trial were both DANA-i and YpsoPump models (we ported the YpsoPump driver we’d written for the Baker trial into their app). They’ve also published some follow-on studies.
The name CREATE apparently came from “Community-deRivEd automATEd insulin delivery”. Maybe a bit of a stretch, but it’s just a name…
The summary “Conclusions” from their initial paper:
In children and adults with type 1 diabetes, the use of an open-source AID system resulted in a significantly higher percentage of time in the target glucose range than the use of a sensor-augmented insulin pump at 24 weeks.
The CREATE results got a lot of press around the world, showing that open source AID can be both safe and effective, and this is also borne out by our study. And the n=1 experience of thousands of “DIY loopers” around the world of course!
What’s next? CloseIt
We now have a new trial starting up, supported by JDRF and this time split across two sites: the Baker Institute in Melbourne and the University of Otago in New Zealand. The last trial didn’t have a “catchy” name, but this time the clinicians came up with CLOSE IT (Closed Loop Open SourcE In Type 1 diabetes). I’ve been writing it as “CloseIt”.
CloseIt is studying the same oref1 algorithm as OpenAPS, AndroidAPS, FreeAPS X, etc. It’s actually packaged within our “Lotus” AID app, which is instrumented with extra logging to facilitate the metrics the researchers want to gather. But the underlying algorithm is the same. The CGM being used is Dexcom G6, and the pumps are the latest generation of research YpsoPumps.
As a really high-level description, this trial will have 75 people spending 3 months using the AID system in Hybrid Closed Loop mode, the traditional mode where we count carbs and bolus for meals. That’s already been studied. But then about half of them will be switched to NOT announcing food for the next 3 months. What many people refer to as “Fully Closed” looping. Note that the participants in this trial will be quite varied, with experience levels ranging from people who haven’t used a pump before through to some users who’ve used other AID systems. Before anyone asks: I’m not involved in the participant selection.
This is the life I’ve been living for the last 2 years: not counting carbs and not bolusing. But as an extension to my own N=1 experiments, this trial has been carefully designed by clinicians and statisticians to produce statistically-significant results. Including observing the differences between NovoRapid and Fiasp insulins.
I’m happy that along with the dry “boring” metrics of clinical data, the protocol also specifies a bunch of psychological measurements. When I switched to fully closed looping I did notice my clinical results “worsened” slightly, but my doctors and myself agreed that they were still “great” and that the psychological benefits were worth a lot. And now we get to measure them in a wider sample set!
It’s of course going to take a while to run everyone through the trial process (they don’t run them all in parallel) and then to analyse the results, so don’t expect the paper from that to be published for a few years. But there should be a lot of data (and experience) coming out of this.
And I’m not staying idle in the meantime! This is just one of a bunch of projects I’m now involved in.
The path forward
Meanwhile, rejoice in the new evidence produced by the Baker AAPS trial, the NZ CREATE trial, and lots of related studies that are gradually publishing around the world.
I’m a strong believer in many of these open source systems. Mind you, I’m obviously biased as someone who’s been using these systems for years and have seen the benefits in my own life.
I look forward to more people getting access to them, and helping build the clinical evidence is one of the things I can help with.