STD Agreement Custom Field

cmob

Hi all,

I'm trying to improve a STD Agreement custom field that we currently have on our system. Currently it only works with processed results and not in the review window prior to processing, which would make system checks much easier.

The custom field is as below:

100*((%.%..SUM(EQ(STDFlag,1)*(Area/Value))))/(%.%..SUM(EQ(STDFlag,1)))/(Area/Value)

Where STDFlag is a simple dropdown custom field on the Sample Set where you mark the working STD.

However, I want a singular custom field to compare an average of the initial six system suitability standards (label SS), with a singular check standard (label c).

My thinking is something like this:

100*(((SS*...AVE((Area)))*(C...(Value)))/((C...AVE((Area)))*(SS*...(Value))))

But I'm not sure if this is the correct way in using STD/Component weights in a calculation.

ydan1977

do you need to use CF for std agreement

Why not name Standard and control which was quantified will give you % deviation of n=6 v control

then simply 100- % deviation as CF if you want to do that

HeatherLongden

thank you ydan1977  that was how I would have answered. This is how most of our users calculate Std Agreement, by leveraging built in tools in Empower

cmob

I've not looked at it from that perspective. I'll see if I can get that to work, is there any literature available on this?

ydan1977

See attached short guide

Empower2018

Thanks for that attachment ydan, its very helpful. But what value does the 100 represent- standard weight or concentration?

cmob

Many thanks as well.

@Empower2018 - the 100 typically represents concentration of the standard. If you would prefer to work in weight, you will have to correct for it in your sample set using dilutions. As long as the approach is consistent across standards and samples, you will see concordance.

@ydan1977 - the proposed 100 - %Deviation looks interesting compared to what we're doing currently. However, if I use the areas and amounts provided for STD A and B with the formula (STD A AVGarea * STD B weight) / (STD B AVGarea * STD A weight) *100 I calculate ~ 99.82% concordance which is a little different compared to the two recoveries you've calculated.

Actually on writing this, it appears the RSD you have for the % Recovery is the STD agreement (100 - RSD = 99.82%). Could you confirm if my understanding of this is correct?

Empower2018

Thanks cmob, also, wouldn't tit be more correct to add an ABS to the custom field formula just for the occasions (albeit rare) where you may get a negative number...ie ABS(100-% Deviation)

ydan1977

% RSD here is pretty meaningless for N=2 , but is part of summary functions from report , not subraction

Indeed ABS would work or expand for CF for criteria user ENUM,GTE LTE and limits 98.0 - 102.0 or range using CConst1-9 where possible

MJS

cmob-
One other thing to note from your original post..."Currently it only works with processed results and not in the review window prior to processing."  The custom field you had used an intersample calculation (the %.%..SUM(EQ(STDFlag,1) part.  That works for processed results because a result must exist for the formula to generate a value.  It's looking for a saved result file in order to add to the existing file as it processes.  When you are simply reviewing data from the sample set prior to processing, there is no result file yet as nothing is saved and so the field can't find anything to use to calculate this value and therefore will never work.  Using the built-in fields and, effectively using your initial injections as standards to generate a calibration curve as the thread's direction has taken is better as you can generate that in Review and then you can easily preview the results as it effectively bypasses the requirement to save result files.

cmob

@MJS - Thank you for the detailed response. I will be looking into the (100 - %Deviation) or (Amount*100/Control Value) solutions. It makes much more sense to stray away from arbitrary custom fields.

Thanks everyone.

Empower2018

Just out of interest how is the % deviation calculated then? Is it the area of each injection of B divided by the average area of the5 injections of A and multiplied by 100? What exactly does that figure represent?

ydan1977

I'm guessing 5 injections used in calibration curve then % recovery for S2 injections used off that ,, I get close to answer but not quite ,, I do not have full precision for areas (documents from Waters training)