Tandem T-Slim Bolus Settings

I don't think the stuff on the page you quoted is aimed at people other than auto-immunes with no residual insulin capability. We (auto-immunes) can, at least in theory, stabilize our BG over 250, though for me I would feel terrible and be looking for something to eat. In practice even for auto-immunes I think it is a dumb algorithm; my correction ratio goes up when I exceed 250, I have "sticky" highs, and, mathematically, the algorithm sucks; it's necessary to measure the half-time (half-life) of the decrease to get a number that is meaningful.

The simple assertion that you can manage DIA and correction factor at the same time is totally, utterly, bogus and wrong, wrong, wrong. (Unless you know how to do it; not the way they imply!)

I used extended boluses on my Omnipod to match its simplistic linear algorithm to different meals. If I eat protein I get an extended bolus, but if it's a snack (not normally a meal) that has a low adsorption rate, like peanuts, I'll extend the bolus too.

I also don't do incremental boluses outside the Omnipod meal-action time (unpublished, undocumented, but I think it is around 15-30 minutes) but significantly before my insulin duration time without disabling the correction. So boluses in the 30-90 minute time frame after a meal are calculated just using the I/C ratio; no correction even if I measure my BG.

I do work round all issues as well ;-) Just keep on testing.

I think you still don't understand.

It's not what the "duration of insulin action" really is; that's not what the pump is asking for. It's what the number is that will persuade the pump to give the correct correction.

Yeah, the wrinkle of changing ratios as a function of BGs occurred to me. And the fact that I'm still making some honeymoon insulin (whether it's immune-moderated LADA or non-immune MODY is still unclear) does complicate things.

I like John Walsh's stuff but some of it is a little too formulaic to me--then again, that's probably the point, i.e. these are formulaic starting points for what really is an art as much as a science.

I don't understand your point about the problem with doing one experiment to determine both DIA and CorrF. If my blood sugar is stable, and I bolus, why wouldn't the results help me infer both DIA (by seeing for how long the insulin pushes my BGs down before they stabilize) and CorrF (by seeing how low those BGs get pushed by that bolus)?

You're right, I still don't understand the distinction you're trying to make. Do John Walsh and Gary Scheiner also seem confused about this? I don't mean that as a snarky rhetorical question.

Walsh explicitly talks about the dangers of using an incorrect DIA to substitute for an incorrect carb/insulin ratio, and in that case, I buy his argument. If insulin actually acts about two to three times as fast in your body as it does for most folks (and don't confuse the shorter "insulin action time" reported on drug documentation with the longer "duration of insulin action," see Pumping Insulin p. 99 for a discussion), then your approach makes sense to me. If it doesn't, then, at least as best I can discern, you're using an incorrect DIA to compensate for an incorrect carb/insulin ratio and perhaps also incorrect basal rates.

But that's just as best as I can discern, and ultimately all that matters is what works for you. I enjoy these discussions because it helps me better understand my own care, and just because I'm intellectually interested in these issues (perhaps in part as a coping mechanism). So by all means feel free to just agree to disagree, and we certainly don't need to keep banging on about the same points.

I reviewed Walsh's argument at the link provided. It is persuasive.

In the last three years or so, I've increased my duration of insulin action from three hours to four and one half hours. What caused me to do that were some serious lows between the four to five-hour post bolus-mark that appeared to come out of nowhere, especially since my IOB was 0.0! While the tail of the insulin curve seems insignificant, it doesn't take much insulin to drive you hypo when the food action is dissipated.

I was reading on one site today that for a 75 kg or 165 pound adult, a 100mg/dl level of blood sugar is equal to 5 grams of glucose, or 1 and 1/4 glucose tabs. Perhaps it is that seemingly insignificant tail of bolus insulin that upsets control and leads us to believe that the problem is a bad insulin to carb ratio or an incorrect basal rate.

Walsh's punch line is hard to refute: "Select your DIA time based on research - DON'T change your DIA to fix control problems!"

So, I've moved my insulin duration on my pump from 4.5 hours to 5 hours. I'll watch and see what happens.

I hear you. My current basal is 0.375 u/hr (9 units/day), and I got there because the doctor started me at 0.4 (12 units/day, which is how much Lantus I was injecting when I was on MDI), and it was clearly a little too much, and 0.375 and 9 seemed like nice round numbers, and it turns out they appear to be pretty bang-on, my BGs stay super flat if I don't eat or inject bolus insulin, and I wake up right where I want to.

Insulin duration does seem fairly easy to actually determine empirically, though. I plan to experiment sometime when I'm stable around 120, and then give a small bolus like 0.5 units, which should drop me to around 95, and use that experiment, run a few times, to really nail down both my DIA and CorrF (with the caveat that they will of course vary a bit based on BGs, time of day, and other confounding factors). I get the sense this isn't something either you (acidrock23) or jbowler would support, though it's not entirely clear to me why.

We're entirely on the same page. I've moved mine up so far from the 4 hours the doctor originally set to 4:30, and after some actual empirical testing I suspect I'll end up moving it to 5 or 5:30.

Some of this argument might seem purely rhetorical but it does make a difference when it comes to trying to learn a lesson from a BG misadventure.

Deciding whether the insulin to carb ratio is off or the basal rate is at fault is important. When the bolus is doing the work of the basal or vice-versa, drawing the right lesson gets muddy. I prefer to learn from my mistakes!

This has been a good discussion.

Agreed! And even if some of the discussions end in disagreement, I like that we can have frank discussions around these issues with each other, I'm a big fan of this community in that regard.

All these ratios and math relationships are good rules of thumb but many times I've been saved by going with my "gut feelings." Running good BGs is definitely a collaboration of science and art!


It's not a straight line that suddenly stops and becomes flat. If you perform a sufficient number of blood tests over the time period you can find out how the line slopes and where it ends up then work back to find both the correction ratio (which is easy) and, only so long as you know how your pump algorithm works, some measure of the insulin action time.

Roberts/Walsh actually describe the issue more eloquently than I can when describing DIA; "how long a bolus takes to bring the glucose down". Down to where? Look at the graph right next to that statement in Figure 3!

John Bowler

> Walsh's punch line is hard to refute:
> "Select your DIA time based on research - DON'T change your DIA to fix control problems!">

I don't understand what that means. I keep reading it and trying to deduce what they mean by "research" and coming to the conclusion, based on the rest of the article, that they mean research performed by medical researchers.

John Bowler

Agreed this requires multiple tests, though a CGM will reduce how many are required. Agreed correction factor should be fairly easy to determine, but just because it's easy doesn't mean it's unimportant!

I don't understand why I need to know anything about a pump algorithm to measure DIA. Heck, I could do the precise same test with a syringe. At time A, when BGs are stable and at least somewhat elevated, inject X unit(s) of insulin (whether via pump or syringe is irrelevant, neither the body nor the insulin can tell the difference). Don't eat, don't exercise, don't do anything else to mess with glucose levels. Monitor BGs to assess when the insulin is no longer lowering BGs at time B (probably 5-6 hours). The time interval between time A and time B is the DIA.

The fact that you're bringing in a pump algorithm suggests to me that you may be overcomplicating this, i.e. your definition of DIA doesn't correspond with others' definition of it. I get the sense that maybe you're using DIA to refer to some sort of extended bolus process, where the pump pushes insulin into the body over an extended period of time. If we had super-fast-acting insulin and/or intravenous delivery, that is how things would operate, and it's how a healthy body does operate.

Does this make more sense? Does it clarify why we seem to be talking past each other?

And the graph (Figure 3) is silly. After BGs level off, they then have them plunge, apparently to zero (assuming that's what the unmarked axes represent). And they want there to be no lows within five hours of the injection, which implies they're okay with a six-hour low, which given that they think most people have DIAs for Novolog/Humalog between about 5.5 and 6.5 hours, makes no sense. Pumping Insulin is better here, there Walsh writes, "DIA time is revealed when your glucose stops falling and becomes a flat line without going low."

Thought it might be nice to see Figure 3.

From John Walsh.

Nope, they mean research *you* do on *yourself*, at least that's how I read it. You need to figure out how long the insulin you use acts in your body. Again, I think you're overcomplicating this.

I think Walsh did mean research done by medical researchers. He also endorsed personal testing as an alternative to discover one's own DIA:

Research studies done 15 years ago accurately measured DIA (referred to as pharmacodynamics) for use in pumps or the new bolus calculators (BC) now becoming available in glucose meters. The results of DIA research studies have never been widely distributed, so people often use an inappropriate DIA setting, often one that is too short. Because the DIA has such an impact on control, always select the DIA time from research study results or from direct testing (see below).

Just came back to check this out and wow, all the responses! I really appreciate all of the feedback here guys. It's very refreshing and nice knowing the people actually care and want to help others (and myself) out. It feels great to be part of such an amazing community here!