Auto calculate QA from BG - let's get it right

97 posts, 13 contributors

Search the DAFNE Online Forums

Simon Site Administrator
Sheffield Teaching Hospitals
560 posts

Theres been discussion and requests for an improved auto calculation of QA in the site and mbi,e apps, which take into account corrective dosages as well, and I'd like to get a consensus on how we should do it.

The first thing I should say is that we don't have the ability to deal with any lawsuits, as we aren't part of the DAFNE program and are a group of graduates just like you who do this in our spare time; so my first diorite is ensuring we dont get sued for providing inaccurate medical advice or similar.

That said , we also want to help graduates in any way possible, so I'd like to go ahead with this function and have created this forum post to ensure we get it right.

So, the auto calculation of insulin dosage matched to CPs is simple - its just take the CP input and multiply by the ratio for that entry type. The tricky bit is how to calculate the corrective dosage. An initial first pass at this would require users to enter how much they expect their BG to lower after injecting 1 unit of insulin. The app could then calculate the difference between input BG and target BG, and use this to calculate the number of corrective units needed. That way it isn't giving medical advice, just crunching some numbers.

However there are a couple of questions in my mind:

1. What would we aim for with the correction? Each entry type has a target range of mmol/l readings, would we aim for the middle of this, or err on the side of caution and aim towards the top end of the range?
2. Would we need to have different 'expected BG lowering values', for different times of day? Or for different levels of high BG? For me I find I can expect my BG to go down by 2 for each unit if. Or receive insulin, but as with everything diabetes related, everybody is different.

Thanks in advance for your help!

Podarcis DAFNE Graduate
Darent Valley Hospital, Dartford
14 posts

1. For the manual corrections that I presently make I use a calculation baseline that is one third of the target range above the lower limit of the target range. So, if my target range is 4.5-7.5 and my BG 11.3, I would add (11.3 - 5.5) / 2.2 = 2.63 (rounded to 3) QA to the app's computed dose. The 5.5 is my baseline figure, one third of the 4.5-7.5 range above the lower limit, and 2.2 is my personal BG:QA estimate, ie 1 unit of QA lowers my BG by a bit more than 2 mmol/l.

I would argue against setting the baseline figure towards the top end of the range for this reason: Everyone likes to be at the lower end of their target range (don't they?) so, if they find that the app will calculate a corrective adjustment that is smaller than the one they would themselves have calculated, they might be tempted to compensate by reducing the BG:QA setting. This would then lead to overly high correction calculations when the BG is well out of target range, and could promote that awful phenomonen of high-low-high-low swings from overcorrection that we all hate so much. We must encourage people to enter their true estimate of their own BG:QA ratio in order to get the best results possible.

But I will accept the halfway point of the target range as a reasonable baseline, and it would be easy to understand if set there. Or it could be a configurable setting, but that is probably overkill.

2. For myself, I use the same BG:QA ratio at all times of the day and no matter how far out of range I am. Other people may be different - please tell us.

novorapidboi26 DAFNE Graduate
NHS Lanarkshire
1,800 posts

This would be a good to pursue, but as you say, everyone is different................

The majority of folk in my experience will have different correction factors according to the time of day [ie. insulin resistance].......

However me, and probably a few others have changing correction factors depending on our BG levels..............

So to suit everyone, some way of entering in these correction factors both by time of day and by BG levels would be good. I suppose just one of these options should be used at a time, although some people may find that they need to consider both scenarios at the same time. But I think that is just getting too we all should be on or near our targets right...........?......... Wink

I personally think the safest choice would be to use 1unit/2.5mmol/l, but maybe upping that to 1unit/2mmol/l might be more successful for some............

As you say, its a risky move, as I know some members here are not in as deep as me and others in terms of their diabetes management........

Maybe a clause in terms and conditions stating that the application calculation function is only to be used as a guide to accompany your own mental arithmetic.....

Podarcis DAFNE Graduate
Darent Valley Hospital, Dartford
14 posts

I would be interested in knowing how you calculate dose corrections for various BG levels. So long as you can specify it there is no reason why the app cannot apply the same logic and do it for you.

novorapidboi26 DAFNE Graduate
NHS Lanarkshire
1,800 posts

I worked it out by trial and error, so record how much your correction dropped you by at different levels. Making sure all other variables are constant is important obviously, so it can take some time as you can imagine........

But yeah once it as been established it should be possible to enter this information in and get a correction dose out........

My current correction factors are:

below 10mmol/l = 1 unit for 2.5mmol/l

10-13mmol/l = 1 unit for 2mmol/l

13-17mmol/l = 1 unit for 1.5mmol/l

above 17mmol/l = 1 unit for 1mmol/l

So a logical approach with the limitations of MDI, I imagine with a pump and extensive trial an error more factors could be established should they need it, but I personally dont need more than this........but I try not get into the higher levels if I can, still do sometimes though.......;)

Podarcis DAFNE Graduate
Darent Valley Hospital, Dartford
14 posts

Thanks for that info, novorapid, that's very helpful. At present for myself I apply 1 unit for a bit more than 2 mmol/l (say 2.2, but it's usually an approx calculation, though with the app doing it, as I hope for, an accurate one will be possible). Like you I also have occasional very highs, especially after sport when I have been stuffing sugar into me to keep from going hypo while playing and then ended up with far too much in my system (36mmol on one occasion not long ago).

Can I ask you, though, do you treat each band separately, or make just one calculation? Eg, if your BG was 20 and your target was 5, would you correct by 15 units (ie 1 unit for each mmol/l, using your rate for adjustments above 17mmol/l, or would you compute the correction in a banded fashion as follows:
correct 20mmol/l to 17 mmol/l with (20-17) / 1 = 3 units
plus correct 17mmol/l to 13 mmol/l with (17-13) / 1.5 = 2.67 units
plus correct 13mmol/l to 10 mmol/l with (13-10) / 2 = 1.5 units
plus correct 10mmol/l to 5 mmol/l with (10-5) / 2.5 = 2 units,
giving a total correction of 3+2.67+1.5+2 = 9 units.

I suppose you must use the former, since the latter banded method is too complex to do in one's head. But maybe the banded method is more likely to give consistent results, and a computerised app could certainly perform such a calculation. That would require more settings data to be stored in the app, though.

Would anyone else like to say how they correct for very high BG? As Simon says above, let's get it right. To do so, we ideally need comments from many users. Do we need banded correction rates, and if so how should they be used?

Stew B DAFNE Graduate
Norfolk and Norwich University Hospital
125 posts

Crikey, my head is spinning..
Presumeably any function which calculates QA and correction values will be optional (and hopefully not the default option)? For me, the beauty of the DAFNE diary is that it lays out the information for me to make a judgement. Yes, much of the time the QA calculation is a straightforward one, but often I have to take into account things like planned exercise, temperature (things are very different for me during cold weather) etc. DAFNE is a peculiar mixture of science and "art", and of course I may be becoming a grumpy old man, resistant to change..

marke Site Administrator
South East Kent PCT
669 posts

make that two grumpy old men Very Happy I think it should be defaulted to off since it will be too complicated for 90% of people and I don't believe its possible to get
it 'right', we are are not machines that can be defined by formulae rather than random BG events looking for a pattern Wink

novorapidboi26 DAFNE Graduate
NHS Lanarkshire
1,800 posts


No, I don't correct using the banded fashion that you set out, just a simple matching up of BG level to the strength of 1 would be madness to try and do your banded version.......

I came to those corrections by simply observing what 1 unit done at each band of BG level......once I got enough readings to spot a pattern I was confident enough to use them...

I think having a QA calculator would be possible, and I see there are concerns as to whether it would be on/off. Ultimately it will only be a guide, a back up to your mental arithmetic. I can never be anything more and so its irrelevant whether its on or off by default....... Very Happy

Phil Maskell DAFNE Graduate
Nottingham University Hospital & QMC
194 posts


From your corrections if you were 17 you would correct 10 or 11 units to get to 6 or 7mmol? For me I would be flaked out at that, shows how different people are Very Happy