This 2014 article published in the Journal of Diabetes Science and Technology written by John Walsh and colleagues makes a pointed distinction between the terms, “duration of insulin action (DIA)” and “insulin action time (IAT).”
I realize that the original poster asked this question in the context of MDI usage but thought many insulin pump users might find this info useful.
The insulin duration times in your table, @Steve_Mann, look more like IAT times rather than DIA times.
It’s taken me some time to understand his point. He contends that DIA is longer than and completely contains IAT. The reason for this is due to the suppression of basal delivery in the test that measures IAT. The DIA time is measured with the usual basal background insulin. Here are two graphs from the study.
Why does this distinction matter? If we, as pump users, input the shorter IAT instead of DIA into the pumps settings then we take on additional risks of hypos due to insulin stacking.
Sorry for the detour into the weeds but I think many pump users set their pump DIA too short and then compensate with other pump settings. This distorts things and makes troubleshooting confusing.
The developers of the Loop DIY algorithm allow choices of various insulin models. The “Walsh” model enables choosing whatever DIA you fancy and that affects how Loop calculates its insulin delivery. (The Loop graph below shows the Walsh Model at a 6-hour DIA. This model, when chosen, permits any time to be selected.) The “Rapid-Acting – Adults” model fixes the DIA at 6 hours and uses an exponential curve to calculate insulin on board in its algorithm. I use this model.
Here’s an explanation from LoopDocs.
Insulin Model
There are four insulin models to choose from; Walsh, Rapid-Acting Adults, Rapid-Acting Children, and Fiasp. If you want to read the nitty-gritty discussion that went into the development of the Rapid-Acting and Fiasp curves (collectively called “exponential insulin models”), you can see that in GitHub here.
We highly recommend selecting one of the exponential insulin models (in other words, not the Walsh model).
A common new Loop user error is to select Walsh model in order to easily shorten their insulin duration (DIA) to one like they used prior to Looping. This almost invariably leads to insulin stacking. If you would like to read more about why the duration of insulin action is important in Loop vs how you’ve traditionally used it, please click here to read a blog post about the subject. In summary, choosing Walsh curve just to shorten your DIA will lead to insulin stacking and less than desired bolusing recommendations.
You can click on each model and see what each model’s insulin activity curve looks like, active one selected in blue.
The differences between the three exponential models (two Rapid-Acting and Fiasp) models has to do with the timing of the peak insulin activity timing. Not surprising, since Fiasp is marketed as the “faster acting” insulin. Currently all the exponential models are defaulted to an insulin duration of 6 hours, but the peak activity of the curves differs:
- Rapid-acting adult curve peaks at 75 minutes
- Rapid-acting child curve peaks at 65 minutes
- Fiasp peaks curve peaks at 55 minutes
The key takeaway here is that setting a reasonable DIA time in your pump does matter. I can remember when I used an “aggressively short” DIA, like three hours and every once in a while getting hit with an unexpected severe hypo that in retrospect could only be attributed by that seemingly innocent “insulin tail” where I had thought that it was not consequential.
As more of us move into automated insulin dosing systems, we should not fall into the trap of carrying over a too-short DIA that may have appeared to work for us with a non-automated pump.
