Insulin duration

What is the best method for testing insulin duration?

I can tell from a no-dinner day that my lantus wears off around 20 hours.

How about fast acting insulin? How do I get a better estimate of my Humalog duration?

1 Like

What I have found works best for us is to look in the evening (because you don’t have any stray IOB at the time), to see any unusual events happening in the tail end. We find, fairly reliably, a sudden change in BG behavior, happening, for us, around 5:20-5:40 after the last bolus. Typically, we see BG suddenly (but slightly) changing its trend, and often going up 5-15 mg/dl before stabilizing again. This can be seen on a CGM but is hard to perceive without one.

Another sign is – when you overdosed a bolus, how long do you see the effect of a sustained bolus-induced low? For us, it never lasts longer than 5:30 minutes.

A third sign is – when you stack multiple injections, how long do you see a stacking effect? This one is harder to read.

This number is different from what we end up dialing in our pump. For convenience, we dial a much shorter number.

I would not have realized that it was so long, had I not been warned that it can easily span more than 4 hours. The major, visible effect of a bolus, for us, does not go beyond 215 minutes. The rest is subtle.

1 Like

With or without CGM ?

This method of checking both your Duration of Insulin Action and your Correction Factor at the same time, from Walsh and Roberts’s Pumping Insulin, of course also applies to injections:

Any time your glucose is reasonably high (usually above 250 mg/dL or 13.9 mmol/L) and it has been 5 or more hours since your last bolus and 3 or more hours since you last ate, give the correction bolus that your pump recommends. Check your glucose every 30 to 60 minutes or track it on your CGM for at least 5 hours to see when the high reading comes down to between 70 to 130 mg/dL (3.9 to 7.2 mmol/L) and remains flat without going low. The time at which your glucose stays flat and falls no further is the DIA time to enter into your pump. For example, if your DIA time is set to 3 hours, you should be at your target glucose within 3 hours and have your glucose fall no further for the next hour.

[EDIT after eating and raising blood sugar]: Of course you needn’t check your correction factor at the same time, and therefore needn’t be “reasonably high.” I don’t see why the test for duration of insulin wouldn’t work if you start out with a target BG.

Also, Walsh and Roberts emphasize that your basal rates have to be correct first, so that basal isn’t helping bolus do its job.

1 Like

The best way is going to be with some periods of not eating, so the basal is all you have going on. And then once you figure out the ballpark, doing a lot of BG testing to get it nailed down correctly.

Since you have an idea of about 20 hours, plan on having no rapid insulin or food anywhere near that time. And also do it on non-running days. Test a lot at that time and watch for the gradual increase of BG.

When trying to figure out rapid duration, you will want to make sure you have your Lantus dose dialed in really well, so you can be sure your BG changes are from the rapid and not the Lantus.

For duration of rapid insulin, keep in mind it is not linear. So that takes a bit of work with eating and injecting and again a lot of testing to see your BG peak and how long it continues to come down after eating. And of course the type of food will have different digestion rates, so that is what makes it tricky. It can also be dose dependent - the amount you take can affect the absorption rate. And also, activity affects your rapid insulin absorption rate more than it affects your basal absorption rate. So all of these things take a bit of time to figure out.

If you have want to test it with a high BG, inject and watch your BG come down and then eventually flatten out. The last portion of dropping and flattening out is your duration.

To make it easier, start with a general ballpark of ~4 hours for Humalog and then adjust your particular duration up or down after testing it and figuring things out a bit. This makes it a little easier if you have a general idea of around where to start.


I disagree with this part. What I see is when the BG goes over 200 that ratios change and effects are different. Life is just different between 200~400 as compared to between 70~180.

That makes more sense to me.

I find that if we only look at the most visible effect, we underestimate DIA. For us, we can see very clearly stacking effect up to 5 hours and more, but the most visible effect of insulin stops around 215 minutes (+/- of course, as usual)

We have 4.5 hrs programmed into our Tandem pump. This was based on watching the curves of the cgm. Previously on the Animas we used 3 hrs but didn’t know enough on the concept or consider it important enough to test.

I think the 4.5 hrs works better as when on the Animas, I found myself constantly shaving down the corrections as I knew it was going to be too much. I don’t have to do that (mostly) now.

However when switching from 3 hrs to 4.5 hrs, we had to make other compensating adjustments. For example, our lunch carb ratio was artificially set for not as much insulin due to there really being insulin left over from breakfast. With the increased insulin action, we can set the carb ratio for lunch more proper and now lunch is better when the carbs vary and we are not just getting lucky that the numbers match up.

But for Dinner and Breakfast, we really didn’t have to make adjustments as there was no close meal prior that the insulin action would have been impacting.


I’ve suspected for some time that my duration setting (4 hrs) was wrong, but this post – primarily the part about corrections being too much – prompted me to do the test rather than just quoting it. Last night, coincidentally, I went above my high-alert level about five hours after last eating and bolusing, so I did a correction based on DIA of 4 hrs and then kept an eye on the CGM. Sure enough, it levelled out at target at 4.5 hrs. So I’ve changed the pump setting and will observe how the next few days go. Better late than never; better do than quote.


I must concur with @Thomas: for some folks, high BG causes insulin resistance, so Walsh’s technique mentioned by @Beacher may be problematic. If my BG is over 160 for more than 45 minutes I start to develop insulin resistance. If it has been over 250 for any significant amount of time it generally takes 3x or 4x my usual correction to bring the BG down, and so to cope I have learned to respond to rising CGM values early and aggressively. If you experience something similar, maybe try Walsh’s technique but with a starting BG of 160 and a correction that you calculate will bring you to 80, and watch what happens.


@bkh You are so correct! I also suffer this same phenomenon.

And Welcome to FUD! Looking forward to more insightful posts from you!