Introduction
Okay. So this is going to be interesting. A really fashionable and fascinating topic, but also kind of sensitive. For this exercise, I'm going to steal some (but definitely not all) from an EDQM document aimed to help specifically on this topic.
What do I think first about doing an Excel spreadsheet validation? Just don't :). Try to choose ANY other way to process your data.
All right, you don't have any other way. There is still a potential loophole: if you don't process GxP raw data to GxP data, you can still call your spreadsheet an inventory or something like that to spare the validation.
If you cannot escape, there are some commercially available applications that handle the spreadsheet in GxP-environment: full audit trail, backup, user roles, etc. Maybe that is going to be your solution! - But this solution is worth it (financially) only if you have many spreadsheets to handle. And you are still going to have a computerized system (the framework application) to validate, maintain, etc.
Let's start: general considerations
OK, you chose to validate and maintain your spreadsheet on your own. I hear you, I have been there :)
To be able to start the validation planning, first let's define what is in our hands.
We can apply the decision tree from a previous post.
First question: is the equipment connected to a higher level system? - NO, so let'go forward.
Second question: how the GxP-data are collected? - the answer is 'electroncally', since you will type your data in the spreadsheet (or you use a BCR, but it's still the same way from validation point of view). So yes, this is considered to be a computerized system.
Third question: How is GxP-data reported (and signed-off)? - Let's say we print out the report (let's revisit this topic later, there may be some other solution). So this is a hybrid computerized system.
But what does this mean?
Preparation: how to build a spreadsheet that can be validated later?
In other words: how can we build up our master data that will support our calculation?
Let's go through what kind of support we need?
1. Overall layout & functionality
First you need to construct your spreadsheet. What are to consider?
- input data (and their format) - make a differentiation with different color where you expect input data
- functions to use
- roundings
- any visuals (graphs, etc.) to use
Before starting the actual preparation for the validation, be sure you have all the calculations you need. Play along with it a bit.
And let me quickly add: this preparation step is basically your URS in your V-model. You lay down what are your requirements.
2. Data validation, input cell's edibility, output cells' protection, sheet protection
Let me not repeat the EDQM guideline on these topic, all recommendations in Chapter 3: 'Good Practices' are a must-have in my eyes.
3. Report layout
Since the GxP-data will be the printed paper, you want to have your report fully readable. Prepare the header and footer also in a way that you have place for your wet signature, pagination and also date & time, if you haven't placed those in some cells.
4. Save!
When you are done with everything above, you own a spreadsheet that is ready for validation.
How to do that? Let's evaluate the options in a later post.
Commentaires