What are the best practices for corrections? I’m MDI (Lantus and Humalog), use Dexcom G6, and was always told to not have corrections within 3 hours of each other. Which doesn’t leave much opportunity to correct if you correct at a meal and misjudged how much insulin is needed. What are your correction guidelines?


What is a correction? I haven’t heard that term. Is that when it seems as though the bolus isn’t sufficient and more insulin is needed?

1 Like

A correction is when your blood glucose is higher than you’d like, and you need more insulin to bring it down. It could be due to a number of reasons, not taking enough insulin for the carbs you ate, changes in your insulin/carb ratio, illness, stress, etc.


I don’t have any clear guidelines, but I certainly don’t wait 3 hours before correcting. When my estimated IOB (I’m a pumper) is unlikely to be able to cover the difference between my BG and my preferred range, I correct. A BG hovering at about 200 mg/dl 2 hours after a meal probably isn’t magically going to turn the corner. Or when my BG is rising too much, it doesn’t make sense to wait 3 hours. I just guess how much insulin is needed to correct my BG. I don’t care if it’s too much. Usually I’m able to prevent going too low. I just keep an eye on my Libre and eat something sugary to stop the decline before my BG gets below range.


I think the corrections within 3 hours guidance is a good starting place for people on MDI without access to a Dexcom. Since you have the Dexcom, I would correct whenever you feel you need to. My son tries to wait until the peak of the last bolus has past before additional corrections, although anytime he is over 250 he rage boluses and eats his way out of the corresponding low. We are trying to get him to stop rage bolusing right before he goes to sleep so we don’t have to get up in the middle of the night to manage it when he doesn’t wake up.


So for Humalog, you’re basically saying corrections should be at least 2 hours apart?

1 Like

I picked up some pointers about corrections and timing of them using CGM trend by reading the Ponder book “Sugar Surfing”. It is worth the read. Corrections can be tuned finer than “wait 3 hours” but it pays to read up on it to get an understanding of the insulin duration and what to watch out for if you stack insulin corrections.

There is also a web site:
Diabetes Education | Sugar Surfing


If the dex graph looks like my BG is running away to the upside, I take more insulin. If a half-hour later it doesn’t look like the trend is moderating, I’ll take more. Again and again until I get a response. After 4 corrections fail to change the trend I’ll take a rage bolus via IM and change the infusion set.

The advice to avoid stacking insulin, or wait 3 hours, is from the old times before CGM, where stacking could cause an unobserved BG crash. With the CGM we are safe from that; we can see the BG plummeting, and based on the remaining IOB we can decide how much glucose to take to prevent an actual hypo.

I’m a big fan of the sugar-surfing style of BG management, in which we take early, frequent small corrections to steer the BG in a good direction, rather than needing enormous corrections widely spaced to solve a BG that went way out of range.


I agree with @bkh . I do the same. This is much easier with a pump, though it can be done with MDI, by taking small injections for the corrections while monitoring your CGM. I don’t have any rule as to how many corrections I will do. I’m on autobolus Loop so it really does most of the correcting for me. I still will need to correct for times when I fail to enter carbs, or don’t enter the correct number of carbs or absorption time properly.


Thanks for the book suggestion! The diabetes educator at the Diabetes Clinic suggested it, but most of her advice has turned out to be like the “wait 3 hours to correct” that I didn’t pay much attention to it. My mistake! So thanks for that recommendation!


Generally yes, 2 hours is ideal, but sometimes my son’s patience isn’t quite that good, especially when he has an aggressive upward trend. Then his treatment looks much like @bkh mentions.


Adding corrective doses of humalog after a meal because the meal contained more carbs than the original estimate is absolutely necessary.

Estimating an error from a BG rise after a meal (“post-prandial blood glucose”) is a total crap shoot; the BG numbers depend on actual carbs and actual protein and actual fat and actual fibre and actual alcohol. It is difficult to impossible to work out anything from the BG until all of those have worked through the system. If I eat Desperate Dan’s cow pie I will be all over the place for 8 hours.

What I do is trust the insulin model on my Omnipod; I know it’s really bad, but it’s better than me, and just do corrections when I feel my BG might be too high. I do ignore anything within about an hour of a meal with significant actual raw carbs, but that isn’t necessary because the insulin model of the Omnipod will zap any correction to zero in the initial hour.

For MDI it is a lot more difficult, particularly if you use the Lantus to cover meal carbs (I think most people do that). No corrections within 3 hours of a meal sounds reasonable. Corrections within 3 hours of a previous correction require head-figuring the insulin-on-board (though maybe some of the apps out there can do it). That should be safe if done consistently (consistency to learn how to get it right) but the IOB calculation has to be done.

It should be easy; record carbs exactly as possible, protein approximately and insulin exactly in an app, have the app calculate IOB, based on the manufacturer data, carb adsorption and protein disposition. Enter the BGs so that app can assess the errors in the carb and protein figures/adsorption rates and suggest minimum and maximum corrections required. Someone needs to write that app.

Other approaches, not requiring either an app or any records, I’m being serious here:

  1. Rage boluses: this really works; take loads of insulin when your BG heads through 250 and doesn’t stop. Carry glucose, lots of glucose.
  2. Keep on correcting. Works fine; when the G6 gives you two down arrows reach for the glucose. Very, very inadvisable without a CGM.
  3. Ignore it. Be cranky, the world loves cranks.

I don’t like (3) myself; I just feel bad. (1) and (2) work fine because it is much, much, easier to correct for low BG than high. With a CGM it is possible to see the slope downwards in addition to the absolute level and that is pretty much what makes it work. Of course this means watching the CGM, or using a CGM app and turning on all the various warnings.


I’m also MDI on Basaglar (same as Lantus) & Humalog and use a G6, but I use Xdrip+ instead of the Dexcom app. Xdrip+ has a nice “insulin on board” curve that it plots when you tell it you bolused that I use to decide if I need more correction. Usually if ~2 hours after a correction my BS is not dropping enough compared to the amount of insulin I have left on board (mainly determined by trial, error, and experience) then I will take more. But my insulin sensitivity seems to be pretty dependent on time of day, amount of exercise/activity, and God-knows-what, so your mileage may vary.


Thanks for your experience, Xdrip+ sounds like it would be very helpful. Is it available for Android?

As far as I know Xdrip+ is only available on Android. Docslotnik and others, as well as archived threads here can help you get started if you want to try switching. It’s not as superior as it used to be since Dexcom has been crippling its abilities, like autorestarting sensors, but (IMHO) it’s still way better than the stock Dexcom software.


I don’t know if @docslotnick still reads this forum though. Haven’t seen him post anything in a long time.

1 Like

xDrip+ only supports Android, and can be found on Github here. You can find installs (the apk file) under the Releases section on the right side of the page, with latest here.

There is a version of xDrip+ for iOS but I’m not familiar with it.




There’s an iPhone version, but you have to build it yourself, Apple doesn’t allow installation of health related apps they don’t approve of, consequently you pretty much need a dev license ($99/year) and sufficient experience, and equipment, to build it. It’s the same as looping with the Omnipod - the RileyLink solution requires a major investment in learning to be a SW developer.

Here’s a FUD thread:

And here’s the app:

Apparently ATM it’s possible to do it without the PITA developer stuff (I do have an Apple developer license, I am an extremely experienced SW developer I do have a Mac I bought especially for the experience but developing SW is something I’ve done all my life and it is worse that diabetes.)

I’ll try the (current) <1min solution. Oops: EDIT: available SOON. Maybe I’ll try ~2min, no I won’t; I’m set up for the ~10min solution and no, it’s OHOALM than 10 minutes.



1 Like