Knapp Test for Vial Testing


I’d like to have your help.

I’m qualifying an EISAI with a collection of 100 vials.

I have 25 vials contaminated with particles, and 75 without particles.
The vials were inspected 10 times in EISAI. In the 25 vials contaminated I have some vials that were rejected less than 7 times (probability < 0.7%).

My questions are:
can I use that vials has rejected, or should I discard them?
to calculate the sum of rejected probability should I sum all the times that the 25 contaminated vials were rejected and divide by 250 (25 vials inspected 10 times each), without rejecting the vials that were rejected less than 7 times?

For me, if the vials have particles (and I know they have), make sense to consider them and sum all the “rejected” vials in the 250 inspections.

But, if we can’t consider vials with a probability < 0.7%…
I’m not sure about what I should do.

Could you please advise and help me?

Thank you in advance!

The Detection Rate Percentage (DR%)

DR% = Number of times machine recjects the via x 100 / Number of times the vial is inspected

Few questions I need to ask before you qualify?
-Have you indentified the particle size detection sensitivity?
-Have you optomized the speed of vials?
-Have you considered Brake position?

Considering 7 times and 0.7% Probability looks no problem.
The whole problem is how you optmize and what particle sizes you detect and at what speed.
Concentrate on those important parameters.

Dear all, I have been working on Brevetti CEA inspecting machines for 25 years, aound 15 years for Brevetti and the remaining leading my own company. I was very much inside the Knapp test validation procedures, therefore I will be pleased to answer to your questions.

Hi Claudio,
Warm welcome to this forum.
We need you expert comments on Knapp test and I writing my points below:

  1. What are common precautions when a person does Knapp test?
    2.What are the common errors that are found and how costly the errors will be after validation and in real time inspections?
  2. How to mitigate errors.
    4.What are the calculations and Satistics involved in this whole test? Kindly explian in brief.
    5.What are your expert suggestions regarding selection of machines for this purpose?

Your comments are highly valued and appreciated.

Best Regards

Dear Durga,
Here I am to try to answer to your questions:
First al all we of course need to assume that we already know and 100% accept the fundamental probabilistic principle, it is also obvious that we base our validation on operators standard results and not on laboratory like procedure. As you know 5 operators inspect the batch
10 times.
1)2)3) Operator must not know the results, number of contaminated ampoules ecc. , therefore a supervisor or somebody else has to write
the results of every single test. A page containing the 250 numbers with 10 dashes on the side of the numbers is the best way to avoid
mistakes in writing the results of the rejected numbers for each run. It is extremely important to determine the inspection velocity, that is
ampoules/hour, on the small formats like 1,2 3, ml it is normally 1000 ampoules/hour, therefore the 250 ampoules should
be inspected in around 15 ', if the time is much longer, results will be different!!!
4) Quality factor of every ampolule (QFA) is calculated dividing Total N of rejection/Number of inspections x 10. In order to calculate the rejection efficiency of the second system (machine) we need to calculate the total summation of the
ampoules with QFA between 7-10, we calculate the same for the machine on the same ampoules and we have the
results, Rejection efficiency = Total sum QFB/Total Sum QFA, if the result equals to 100, it means that machine is working with the same accuracy.
I hope I was clear enough, even though I was doing very short what sometimes takes very long to explain, regarding the
machine to choose as you might know there are nowadays many companies making them, I also sell some second hand Brevetti machine which are still reliable.
Let me know if you need more details about the whole fact.
Best Regards
Claudio Milan

Thanks a million Claudio.

Hi Durga, which machine are you using right now? I have been to Copenhagen a few times long time ago, I worked in Lunbeck, do you know that company? Regards

Hi Claudio,
I worked in Pharmaceutical industries in Sterile Dosage forms and Biotechnology for 14 years. Since last year I moved into Kneat Solutions Ltd,. Iam a GMP specialist and we Develop and sell Manufacturing, Validation and Project – Documentation systems for Pharma, Biotech, Medical device and Food (Pharmaceutical foods) Industries.
I worked on Breveti Machine.
Now I moved into quality area. We support our customers with Quality solutions.

Hi Durga, so which kind of Brevetti did you work with, and are you still dealing with them? Please let me know. About the Knapp test did you get a complete description from Brevetti, there is a brochure explaining details…

Iam working in area Manufacturing and Validation IT services. I do not work for any pharmaceutical or biotech company now. I worked on old model ATM18/S. I think I quoted the model right. We purchased a second hand machine and it worked perfect.You explianed all details well and perfect. Thanks a million again.

Dear Durga, as you maybe know I perform maintenance and validation especially on these old machines , ATM 18 and ATM 32, as I lefft Brevetti in 1999, already a very long time. If you know any Company or hear about somebody using that kind of machines, please let me know and I ill be very gratefull .
As well as you have any other question about validation you can ask me .


Hello everybody
We plan to buy a new full automatic inspection machine and we have to qualify and validate it later. I know the procedure of Knapp-Testing but it is very complex and time intensive. The time for having a running machine is very limeted. So does anybody know an other option to qualify and validate an full automatic inspection machine? Maybe with circle testruns and particles with defined particle size?
Please let me know your experience

You can test with defined particle sizes.
But another option in my opinion is very bad.
This is not to offend you. But the fact of particulate matter remains same if you are time to run the machine for a limited time or limited ampoules or vials or a limited quantity.
It all boils to the particulate matter in injectables.So you need to go with standard reguatory methodology rather than skip test.

Hello Durge
I know of no regulatory requirement for Knapp-Kushner. It is not written in the USP else in the Ph.Eur. yes - it is the the most taken method to qualify and validate an automatic inspection machine but I don’t want to believe that it is the only way.
Knapp-Kushner have set no limits for the particle sizes. Yes he gave a statistical overview of the possibility to detect but for the test it self they give no limits.
Other point - for qualification you say the machine must be better or equal than the human eye. In routine production you have to verify the result of the inspection. This is most done by manual inspection. So human eye is supervison, or?
I don’t say that the Knapp-Method is not good but it is extremly time intensive and I just search for alternatives.

Well on regulatory front I totally agree with you that a guideline of this sort or compendial texts are not avialable.
But… we all know for sure regulatory authorities look into technical procedures and reports and or studies guided by Parenteral Drug association, USA.
They will keenly look into such aspects.
During years 1993-1995 (some where in between these years) a paper about Knapp test was published in PDA journal.
Its good that you refer that and find out.
Much of explanation was given on that.

Dear Kalyan !

How can you justify 83 or 95% ?
The detection rate depends on the size of the particles. Normally you reach a diagram particle size with the corresponding detection rate or detection probability. A detection rate of 0.83 means that you can find a specific particle size with 83 % probability.


KNAAP test is a KIT prepared for Automated Inspection System Validation. This KIT is prepared by Manual Inspection. This kit contains specific number of vials / ampoules.

suppose a Knaap kit contains 300 vials.
From these 300 vials: 200 manually inspected good vials and 80 manually rejected vials with glass particle, fiber or black particle and same should be recorded in a sheet and 10 vials with low fill volume & 10 with high fill volume. After this KIT should be verified by number of manual inspector and then KIT will be finalized. Each container should have a number i.e. 1 - 300 and whether it is good or bad(what type of rejection) should be mentioned in sheet.

After that manually feed the data to automated inspection machine according to number of this KIT. Arrange the KIT according number of feed data to automated inspection machine. Run it a specific number of times like 5-10 times. Calculate the efficiency with manual inspection. It should be more than 100% than manual inspection.


good post only thing wrong.
you need atleast 5 inspectors to inspect a product ten times not 3

We do produce contact lens ( medical device), and currently we have a automated lens inspection system to repalce the manual inspection .Is the method can be similar with KNAAP test. As i know, at the automated inspection system, have 13 defects. how am i going to validate all the 13 defects in a row. Please advice

Very difficult question to answer.
The reason is it is built for vials/ampoules that are made to rotate with certain solution or drug product or powder/pellet at certain speed under specified wave length of light against various backgrounds(Black/white/Grey).
It will be real good idea that you contact Parenteral Drug association directly. (

There will be interest groups. Optical Inspection group will be responsible. You need to write them a mail or contact by calling them.