Manual differential correlation between technologists

[milesd3 5]

I've been trying to figure out how to correlate the technologists differential with each other.  There should be some sort of agreement but I cannot find anywhere on the net what is acceptable and what isn't.  the recent diff that we all did the techs neutrophil counts agree right at 15% but the lymphocyte count the difference is something like 32%.  I'm not concerned with monos on down as much but this patient had a high eos count which may be throwing things off somewhat.  .  Is there a better way of doing this?  is this a requirement for JCAHO even?  What is everyone else doing?

 

 

[nziegler 34]

For anyone else that may be interested, this is the Rumke Table.  The easiest way to correlate differentials that I have found.

Rumke Table.pdf

[mrdth5 14]

We use a WBC simulator offered though Media Lab. It allows the techs to test their competency against a variety of different cell types against the experts.  Once a diff is completed, the experts discuss each cell type and identify characteristics of why they called it XYZ cell.  The techs love it and they can do it whenever they have down time.  The website will let you try a demo version if you ask.

[AndreaT 10]

Has anyone created the Rumke Table in an EXCEL TEMPLATE?

[Auntie-D 362]

We use Horiba's QSP - we have two techs on the film bench and both are required to do two slides each day they are on the bench.

 

I've attached one of the reports generated for me so you can see the sort of information the report covers.

 

The program itself is drag-and-drop to allocate the cells into the categories. 

 

All in all it's quite a neat program

LK_Aout 2014_05 Aout 2014_assistant.pdf

[gilmanch 0]

This is something I have been working on also. Few things I have implemented:

 

1. Prodedure that is detailed about what we call for morphology and cell identifcation (we were having variance in bands vs. segs.)

2. Techs signed up for the Media Lab

3. Each morning I or my lead tech pull 3 slides (one each shift) and perform a diff to check results reported. criteria I use is need to agree 10% if results over 30 and need to agree 50% if results under 30. I got this idea and criteria from another lab's supervisor. Since then I have seen other things that need to be standardized more

4. This article has information that I am going to use one day to help standardize us a little more further

 

MORPHOLOGY PAPER.pdf

[milesd3 5]

Thanks I used this on the comparison studies and in the most part the techs did ok.  Well in relation to each other they did, however, the correlation with tech vs instrument 4 were outside Rumke limits for Monos. Do you check the tech against each other or the instrument.  If there are failures what is the next step...?  I didn't provide the instrument diff on purpose  should I have? 

[milesd3 5]

Thanks I will look at this when I get the chance.  I like the idea of some feedback from professionals.  we are a small rural hospital and none of us are far from experts..

[milesd3 5]

This also looks very impressive.  I'll take a look at this as well  thanks

[milesd3 5]

The % thing wouldn't work.  When I was initially trying to figure something out (and why the problem seems so overwhelming) The Neutrophil % difference was at 15 but the lymphocyte % difference was more like 32.  That is comparing 15 techs.  I might have picked out a slide I shouldn't have.  Eos were around 16..?

[gilmanch 0]

Milesd3, what analyzer do you have?

 

something I now have my techs doing is using the analyzer for comparison. They must match the anaylzer somewhat, (i use the same criteria as the man diff) and if they do not, repeat the diff. If they get the samething, then I want them to look at the flags and scatter plots to see if what they are getting make sense. if not, remake the smear b/c something could be off on smear (prep or stain) if the flags ans scatter plot match what could be the variance then they can release it. We have an Advia, and most of our unmatched man diff compared to the auto diff is if the patient is myleoperoxidase deficient, will mess with the diff, give us more monos then what is really on the smear.

 

 

[milesd3 5]

Gilmanch we use a Sysmex xt2000i.  we had the Advia and while it does a great diff when its running ours was down more often then I like to remember .. Besides all techs comparing diffs with one patient slide twice a year, we are also having two techs performing a 200 count diffs each daily (or its supposed to be daily anyway) as suggested by the last JCAHO surveyor which accomplishes two things. Checks the manual diff to the instrument as well as checks stain quality.  I like that feature of the Advia by the way.. I'd seen a few MPO patients with the celldyne but there wasn't any flag to warn the tech  ..

[Mari 0]

I have been researching this topic and came across this resource that offers an exel template for a manual vs automated differential comparison using the Rümke table.  It looks pretty accurate.  I found one evaluation that didn't match exactly, but it was close.   

http://resources.psmile.org/resources/equipment/validation/smile-validation-guidelines/hematology/rumke-differential-evaluation-tool/view

[nziegler 34]

that spreadsheet looks great!  I'm definitely going to give it a try.  the only thing I see that I don't like is that it seems to use the manual count as the "reference" method to do the math... with today's technology, I would consider the automated count more accurate than manual (obviously excluding leukemic/dysplastic scenarios)