False positive qa statistics

[jalomahe]

We want to track our "FALSE POSITIVE" antibody screen rate as part of our QA Statistics. Currently our definition is the Antibody Screen is POSITIVE but the Antibody Identification shows no specificity or is negative.

Does anyone else track this data? If so.......

What is your definition of false positive?

What antibody screen/antibody id methodology are you using?

What type of percentages are you seeing?

I'd like to come up with a benchmark to use for management as to when we are statistically too high. With just three months data we currently show an average of 1.0% that fit the above criteria.

Thanks for your input.

[goodchild]

We performed 1003 antibody identifications during 2012. 70/1003 were identified only as "Inconclusive". We use the Ortho ProVue for screens and the Ortho manual gel method for antibody identification. We have immucor 3% panels we convert to 0.8% for selected cells.

- - - Updated - - -

We performed 1003 antibody identifications during 2012. 70/1003 were identified only as "Inconclusive". We use the Ortho ProVue for screens and the Ortho manual gel method for antibody identification. We have immucor 3% panels we convert to 0.8% for selected cells.

[David Saikin]

I think that just because you run a panel that is either negative or inconclusive you should not consider your screen to be a false positive. It is important to remember that you are working with biological moieties here and not something that possesses reproducible qualities. Before you call your screening results false I think you should repeat the screen (or maybe increase the sensitivity of your ab identification procedure). If it is still reacting then I would deem it to be a valid reaction. An good example is the detection of antenatal RhIg at delivery. Many times the reactions (in gel) are 1+ . . . an abid ends up being negative but I know my screen results are valid . . . What are you going to do with your "threshold"? Sounds like making busy work to me . . . you are still going to have to perform some type of workup to resolve your pos absc. This is also true for those folks (including myself) who use gel and get reactive results with the Ortho screening cells. I prefer not to run Ortho panels and on occasion my 3% altered to 0.8% panels will be negative. If/when that happens I either enzyme pretreat those cells or run an Ortho panel. Usually some reactivity is expressed (anti-E reacting with only homozygous cells - and the pt is E neg -comes to mind 'cuz it happens fairly frequently). To sum up, I think it is more of a waste of time to track this stuff (to what end?) than it is to work it up. If you think your systems are too sensitive then you should consider less sensitive methodologies (which is something the bean counters may recommend depending on your "threshold").

Edited April 11, 2013 by David Saikin
poor spelling

[dmpollock]

To Goodchild:

An unrelated question:

I assume you have a busy transfusion service to do 1003 panels in a year. How many provues do you have to manage the workload? How many RBC units do you issue in a year.

I am curious because our management is considering switching from Immucor to Ortho (which we really don't want to do).

[goodchild]

To Goodchild:

An unrelated question:

I assume you have a busy transfusion service to do 1003 panels in a year. How many provues do you have to manage the workload? How many RBC units do you issue in a year.

I am curious because our management is considering switching from Immucor to Ortho (which we really don't want to do).

I am in error. The 1003 number was the number of individual antibodies identified during that time period, I was looking at the wrong column in my spreadsheet. I'm kind of sad that I made my own spreadsheet too confusing for me to look at only a few months later but that aside the actual number of antibody identifications was 714.

We have one ProVue currently, which took us a long time to get. We're trying to get a second one, but I don't see that happening any time soon. We transfuse >400 RBCs/month. I am more a fan of the immucor panels and the cell selections you get but I love the workflow ease of gel panels from Ortho. We get the Resolve A and B every month but the B panel is for the most part, worthless.

[SMILLER]

I would also wonder at the usefulness of this kind of QA project. If the numbers go up or down, what difference will it make in how you manage your patients? Or is this just another statistic to report to the transfusion committee once a month?

Scott

[John C. Staley]

I'm with Scott, what are you planning on doing with the data? What criteria will you use to determine "acceptable" from "unacceptable" or are you doing it just to make someone think you are doing something useful. I've had to do a few of those over the years and I see this one being one of those. I also agree with David concerning the use of the term false positive in this case. Biological systems are wonderful as well as full of wonder but they never seem to read the book on how they are supposed to act all the time. :abduction

[goodchild]

I have no idea why the OP is looking at this, but if the last two posts were directed towards me: I'm not tracking "false positives" as the OP describes, but when I saw his post I did have some information available with our "inconclusive" category that he might have been able to find a parallel in, so I posted.

I'm looking at this information for a couple reasons, which I'm sure would be scrutinized just as harshly so I'll keep silent!

[John C. Staley]

I have no idea why the OP is looking at this, but if the last two posts were directed towards me: I'm not tracking "false positives" as the OP describes, but when I saw his post I did have some information available with our "inconclusive" category that he might have been able to find a parallel in, so I posted.

I'm looking at this information for a couple reasons, which I'm sure would be scrutinized just as harshly so I'll keep silent!

Goodchild, my post was in response to the original post by JALOMAHE. My questions were simply questions I feel everyone should ask themselves when they are planning a QA a project/study. All too often we simply climb on the hamster treadmill with no goal beyond gathering data to impress some one. (Been there, done that!) Personally I would like to know why you are looking at the information because others may be missing out on collecting and evaluating valuable information simply because we never thought of it. I really hope your skin is not so thin that a couple of questions to provide understanding does not force you into the back ground. Harsh scrutiny is never my intention. :disbelief

[jalomahe]

Thanks to all who have responded to this post. The information is invaluable.

My query was based on a request from my new lab director who thought this was relevent data we should be collecting. Her background in blood banking unfortunately is little to none. Although I have tried to explain blood bank concepts....no one method....sensitivity vs specificity....and just because it's not identifiable (at the time) doesn't mean that there's nothing there.... the director wanted to hear what others are doing.

Basically I was hoping for exactly the responses I received, that this is not a good way to spend time and effort. So thank you all for giving me the ammunition to shoot down this project.

[goodchild]

My comment was written with tongue in cheek, but I failed perhaps to include a smiley face to eliminate ambiguity:p. I really don't know if anyone else would really see the value but for me the information is very easy to compile, so why not have it available? Our LIS has a report I've found that will pull every antibody identification, patient MR# /account#, date/time and technologist ID. I run this report through a macro in excel and voila.

We switched from manual gel to Ortho ProVue gel recently. I like the idea of being able to have a comparison of ABIDs from pre/post ProVue, maybe one day write something. The general information for our antibody ID trends will also be useful for tracking reagents usage for antibody identification.

[Mabel Adams]

Because of the problems with "gel junk" reactions, I can see someone maybe wanting to keep an eye on it like Goodchild--especially if it is not much work to gather the data.  I agree that it not a useful statistic and can hardly be reasonably evaluated by anyone outside of BB who doesn't have a "feel" for the work.