do I need to explain why integration manully by GMP/GLP?

[flora0975]

Does GMP or GLP require chemist explain why he/she uses manual integration instead of auto integration?

I used to use software auto integration first, and then check if every chromatography integration consistence to each other. If not, I usually manually adjust the integration . I did this way in my previous companies.

Now I am in a new company, my boss asked me to explain why I need to do integration adjustment manually every time when I need to. Is this required by GLP or MP?

[tom jupille]

There is no "requirement" one way or another; it is a matter of the company SOPs (Standard Operating Procedure). That said, manual integration is usually frowned upon in a cGMP environment, because it provides an opportunity for "cheating" on specifications. In many companies, the SOPs allow manual integration only in the case of obvious problems, and even then, require written approval from a supervisor.

Bottom line: whatever the company's SOP says should be followed.

[DR]

^^ I agree w/ Tom, but want to add that for many chromatographers who often feel the need to manually integrate, a careful review of basic processing method parameters is usually in order. Getting peak widths, sampling rates, and threshold values just right usually either obviates need of manual integration or points out the need for a better separation (or at least column clean-up/replacement).

[unmgvar]

these days unless special cases it is true that a decent software should be capable of handling almost anything in terms of integration.
since like Tom says manual int. looks like a solution to "hide" things then for sack of traceability and good practice, many will agree that explaining your action is a must.
if you do this too much and specifically for a given application, then you should maybe look and see how to improve the problem you have that forces you to use the manual integration.

[JGK]

Going back a few years when I was managing a chem dept in a GLP environment, the requirement (for bioanalytical work)was that any manually integrated/reprocessed data had to be justified by the analyst and signed of by the Manager before the data could be used.

[adam]

It is true that some frown on it in the industry. I have always found it interesting that many people think the computer will always get it right. Of course the human brain is the ultimate decision maker.

Therefore, in my opinion, manual integration should be OK, and you don't need an explanation generally. But I would be especially careful in a situation where you are close to the spec (as mentioned above). In situations like that an explanation may be in order if you need to use manual integration. For example, if your result is 109.6 and it looks like you tightened the baseline to bring the result below 110; that could be a bad situation.

[tom jupille]

Code: Select all

the computer will always get it right

Actually, from a regulatory perspective, the advantage to the computer is not that it's right (I agree with you that often it's not), but that it's consistent and it leaves an audit trail.

[lmh]

And the computer doesn't know what the target value is (even unconsciously) and has no vested interest in meeting a target value.

[Alex Buske]

[quote="tom jupille"][code]the computer will always get it right[/code] Actually, from a regulatory perspective, the advantage to the computer is not that it's right (I agree with you that often it's not), but that it's consistent and it leaves an audit trail.[/quote]

I presume that manual integration is also audit trailed. Automated integration is consistent in always applying the same integration parameters. In different chromatograms that does not necessarily result in consistent integration. in our lab automatic integration is strongly preferred, each manually integrated chromatogram has to be reviewed by someone else

[yangz00g]

Agree with Tom and Alex.

Been working in a GMP manufacture for over two years, begin to realize consistency is way more important than anything else. In may lab, in general manual intergration is not allowed, except in very rare situation.

[HW Mueller]

Accuracy is not so important? That´s good manufacturing?

[yangz00g]

Accuracy is not so important? That´s good manufacturing?

it's based on the assumpation that accuray/precision are acceptable to the preset specifications.

[JGK]

In the early days of computerized systems which allowed us to amend the integration start and end points to provide a peak area. (long before the days of built in audit trails showing who did what.

We had and SOP which stated that the reintegration had to be done by 2 analysts independently (without the other present) and the results submitted to the manager. The integrated values had to be within 3% of each other to be approved.