How Many samples to Validate a New Method

[Diane]

I am trying to advise a very small hospital in our area that wants to make a change in their compatibilty testing technique (switching from serum to plasma).

They probably do 0-2 crossmatches a day. Does anyone know a required # of specimens that they should run to validate the switch?

[Lcsmrz]

At that volume, why bother changing ?????

I'm assuming that the Antibody Screen would also change to plasma and that we're not talking about a change in technology, such as from tubes to gel. I can't think of any required number of validation specimens, unless you get into automation.

Because their volume is so low, a majority of the validation documentation could come from available research proving that the two methods are equivalent and detailing the strengths/weaknesses of the two and the reason for the change. You could also show that the new plasma tube is FDA-approved for BB work and use any training samples as part of the validation. Training for phlebs, tech and (if drawing blood) nurses needs to be documented, and SOP changes approved in advance.

If the only change is the switch from serum to plasma, I would have my techs read the documentation I provided, maybe take a quiz, and perform one or two parallel tests drawn from the same person, just to see what the new tubes feel like and to document that they've been through it at least once.

[conwaysbb]

When switching to a new reagent or methodology we utilize this as our method of validation:

Evaluating/Validating New Materials

1. Any new procedure or test system should be demonstrated to obtain comparable results to the process being replaced in terms of accuracy and reliability.

2. A test system for antibody detection should be evaluated using a minimum of 30 patient samples, at least 10 of which contain unexpected antibodies, and 20 samples that do not contain unexpected antibodies.

3. The patient samples should be tested using the new system in parallel with the existing system.

4. Determine the number of true positives, number of true negatives, number of false positives and number of false negatives and calculate the specificity and sensitivity of both the new and old test system.

5. An assessment should be made of the test system’s sensitivity and specificity.

CALCULATIONS:

# of true positive results X 100 = % Sensitivity

# of true positive results + # of false negative results

# of true negative results X 100 = % Specificity

# of true negative results + # of false positive results

Example

In a test system with 30 patient samples consisting of 10 samples with unexpected antibodies and 20 samples with no unexpected antibodies the results of of one test system are as follows: 9 true positives, 1 false negative and 20 true negatives

9 X 100 = 90% Sensitivity

9 + 1

20 X 100 = 100% Specificity

20 + 0

INTERPRETATION OF RESULTS:

The minimum sensitivity and specificity of each test system is > 90. %. Test systems that do not meet the minimum requirements shall not be used.

I would assume that you would need to do both validation of immediate spin x-matches for the detection of ABO incompatibilities and coombs x-matching for clinically significant antibodies.. If you utilize any other x-match technique I would also validate.

However, as the previous post stated, why change when they do only 1-2 x-matches a day?

[John C. Staley]

Unless you are using a technique very unique to your setting (2 day room temp incubation in 65% albumin or some such) and half the country is already doing what you want to do, why all the extensive validation? The reagent manufacturer says you can use plasma and the FDA has given it's blessing to the reagent. Run a few parallel tests to see if there is any visible change in what you are used to seeing. Why do more? Obviously patients are not dying right and left all over the country where facilities have already switched to plasma and been using it for years.

[jssa1891]

I think it is a good thing to remain current, no matter how many crossmatches a day the lab does. Moving from serum to plasma may be helpful in avoiding a number of problems such as samples that won't clot and false positives due to incompletely clotted serum. As far as parallel testing, this is pretty much standard of practice in our area, when changing methods, no matter how big or small the hospital. It is a sign of a good quality program and worth the effort. It might also help avoid the lab encountering any surprises after they make the change. What kind of surprises? That is what parallel testing should tell you.

[johna]

I agree with John. What's the point in validating a procedure that's fully approved by all the regulatory agencies and is being used by well over half the blood banks in the country. The results of a validation may look nice on paper for an inspector but would have no practical value. In these hard economic times it would seem more sensible to devote the time spent in doing this evaluation to something more productive and constructive.