Correct GLP for carryover in a sequence run

[unmgvar]

i recently came across a behaviour toward carryover that raised some questions in my mind, and i had to ask the forum for opinions.

in a pharmaceutical laboratory they are having carryovers problems for some impurity applications with one of their type of HPLC instruments.

the two groups in the lab are dealing with the problem in totally different ways:

group A's attitude is to rely on the basic washing capability of the instrument only in order to try and resolve the problem. using the advanced commands is not GLP to them. if the carryover occurs then they simply inject blanks/diluents, one after the other (for hours sometimes) until they get a clean chromatogram. then they start the sequence in hope that the carryover will not reappear by the end of the work. they inject blank/diluent along the work to see if the carryover occurs and act accordingly for calculations. they have a lot of them relatively.

Group B's attitude is to use the enhanced washing commmands available by the software and instrument in order to prevent the carryover.
they insert those commands in between the sequence runs. they also add blank/diluent injections along the work but in their case they have almost no carryover problems.
yet for some reason they hide it!? they print out a sequence table without the added commands for repport and then they add them and start the work!? they also believe that what they are doing is wrong according to GLP.

i believe that both groups are not behaving in accordance to GLP, since both attitudes are the result of fear of audit comment.

Group B's attitude is the worst for they are hiding information, which is even worst then not doing nothing, and the worst part is that in my view they are taking such a risk for doing something that doesn't needs hiding.

i see nothing wrong in adding autosampler wash commands as part of the sequence run and showing that such a procedure is being taken to prevent carryover.

but maybe i am wrong.
any thoughts on the matter?

[tom jupille]

I think that, from a regulatory perspective, GLP or cGMP are more about consistency than about correctness ("document what you do and do what you document").

If the two groups in question are part of the same facility (company or laboratory) and they are handling a problem in different ways, it's time to write (and enforce) an SOP.

The risk in writing a too-rigid SOP is that it may stifle innovation and improvement (by way of example, the US state of Michigan only recently (2003) repealed a law that required all automobiles to be preceded by someone walking front of them and swinging a lantern). It may make sense to simply specify a test for carryover and then leave it to the individual groups to prevent it.

[HW Mueller]

Reading some of the contributions one must come to the conclusion that it´s a good thing that there seems to be some truth to the statement: If you go to the doctor your illness will be gone in two weeks, if you don´t it will take 14 days.

[B.Laudrup]

cGMP/GLP should involve good sensible science (I believe good laboratory practice should not be feared as a regulatory burden). Neither approach described is sensible but I doubt if either would get a FDA 483 if admitted. Good sense says use the programming function because it makes your life easier and gives better quicker repeatable data. If you are washing the needle post injection (the usual way of setting these things up) how can it affect the analytical data! However think about what the wash is doing. I know of a company who were asked to carry out a HCl needle wash but unwittingly started to inject small amounts of HCl down their column because the wash program was not set up properly.

[unmgvar]

Sorry for the delay,
i was too busy until now,

i believe that the fears i the lab were due to a lack of knowledge in the basic need for the regulaty measures.

GLP, GMP is so that we can answer the basic 5 questions:

Why?
How?
Whom?
When?
Where?

so that we can have tracebility, credibility and reliabilty of our procedures.

B.Laudrup

did, in the exemple you mention, the lab encounter some sort of System suitability problems with that application? if so then they were still OK. they had a procedure to monitor their actions. they must of had since you found they were doing something wrong.

that what i am driving at. if you give reasonable answers to the risks you assessed might happen then all is OK. and even if you did not because nobody is perfect you still have the possibility for corrective action.