... the fun part of Medical Device Process Validation!
Great, you finished the Installation Qualification (IQ) and want to start with the fun part?
The Operational Qualification (OQ) is the exciting part of Process Validation – we finally get to produce some parts and see the process performance.
This post talks about:
Why Should You Perform an Operational Qualification (OQ)?
The Operational Qualification has two objectives:
Identify critical parameters, characterize the process, and establish an operating window (also referred to as a process window).
Run the process at worst-case conditions to demonstrate these conditions result in an acceptable product; this further demonstrates that the process is robust against variation of process parameters [1].
Unfortunately, most often, an Operational Qualification (OQ) is performed due to regulatory requirements. ISO 13485 requires a process to be validated if its “resulting output cannot be or is not fully verified by subsequent monitoring or measurement”. This is the requirement about when to perform process validation.
Now, you might ask, “Well, who says that a process validation needs to have an operational qualification?”.
It is good to bring up these questions – always doubt everything.
Unfortunately, there are guidance documents and FDA material out there that explain what a process validation includes. A popular example of such guidance is the GHTF (Global Harmonization Task Force) process validation guidance from 2004 (!). Another resource is the FDA’s presentation by Joseph Tartal from 2015. A couple of warning letters from the agency go to show how serious this really is [FDA Warning Letter #598171] or [FDA Warning Letter #681977].
What is an Operational Qualification (OQ)?
The GHTF (Global Harmonization Task Force) defines Operational Qualification as:
“establishing by objective evidence process control limits and action levels which result in product that meets all predetermined requirements.” [2]
An Operational Qualification is the worst-case challenge of the process at the limit of the process parameters.
That means we must set our parameters to their limits and run the process.
However, we do not even have process parameters at this stage, nor do we know their limits.
This will soon lead us to the next step, but first, I want to make sure that you understand that there are not only always 2 worst cases. While 2 worst cases might be true for easy sealing processes, it might be totally different for a complex injection molding process. We always need to figure out how many worst cases there are and then need to test those during the operational qualification.
How to Establish a Process Window
We already know that the first objective is to characterize the process and establish an operating window. The keyword here is another three-letter acronym, DoE – Design of Experiment.
Design of experiment is a vast topic, but we will try to break it down into a couple of simple steps by using it on a process most of us are familiar with – a heat-sealing process of a sterile barrier system.
The first step of a DoE is to choose the input factors and our output responses.
In the case of the heat-sealing process, the input factors are temperature (T), time (t), and pressure (p).
In some cases, the time (t) is given as speed (v) and the pressure (p) in force (F), but for the sake of this example, we will stay with T, t, and p as our input factors.
Let's assume our output response is seal strength (Fs).
Based on this information, we can already say there are 23 cases we have to produce and analyze to get a better picture of the process.
You might ask where the 23 cases are coming from. The 3 input parameters are the exponent, and the basis 2 is for each end of a specific parameter, i.e., temperature high/low, time high/low, and pressure high/low.
In the next step, we combine these extremes and assign "1" to high values and "0" to low values.
Table 1 Full Factorial Design for 3 Inputs
Table 1 shows the different combinations of the input factors. Each combination will be produced and measured for its seal strength (Fs).
NOTE: TBD stands for "to be determined".
The output response is then fed into a statistical software like Minitab to analyze the factorial design.
A complete and more comprehensive guide to designing an experiment can be found here.
While a heat-sealing process is manageable regarding the number of cases to test, a more complicated process like injection molding, where the number of input factors can easily go into hundreds, it is far more difficult to characterize the process like this.
Even though Minitab helps you analyze the design, you might still be asking, "What values do I choose for all those 0s and 1s?"
Well, that is a good question.
The material datasheet can be a good and valuable source of information to answer this question. Some material suppliers provide information on where their products work best.
The above process parameters are the apparent considerations when one thinks about Operational Qualification and worst-case testing. However, other aspects need to be considered too; examples are:
Software parameters
Raw material specification
Process operating procedure
Material handling requirements
Process change control
Training
Environmental conditions
Operational Qualification (OQ): Running at Worst-Case Settings
Assuming we found the following set of process parameters:
Temperature (T): 150,0±10,0°C
Time (t): 2,0±0,5sec
Pressure (p): 3,0±0,5bar
The process parameters set on the machine for the two worst cases (upper and lower) are shown in Table 2.
Table 2 Worst-Case Parameter Settings
Now you know what parameters you must set to run the Operational Qualification studies.
To answer the question of how many samples you must produce, read our blog posts about "Process Validation Sample Size" and “Statistical Tolerance Interval”.
A tip related to sample size and Operational Qualification is that you can use reduced reliability levels (confidence level remains at least 95%) for OQ runs.
These lower requirements for OQ are because these tests are performed under worst-case conditions, which makes them a stress tests [1].
While the above sealing process example might be easy, we can briefly touch on something more difficult, like a cleaning process. In a cleaning validation, a worst-case scenario could involve selecting the most difficult-to-clean product residue (e.g., highly viscous or sticky substances) from the most difficult-to-clean product. The most difficult-to-clean product might have various product features, like threads or intentionally rough surfaces with undercuts. You see that an operational qualification can very quickly become very big (and powerful).
What Do We Mean by "Worst-Case"?
Worst-case conditions are the settings for the input factors that cause the worst-case performance of the output responses.
In the previous example of a heat-sealing process, a lower limit setting might correspond to undersealing and vice versa.
Author: Simon Föger
Are you unsure how to perform an Operational Qualification (OQ)?
Contact us today at office@sifo-medical.com, and we'll support you perform an OQ correctly.
Further helpful links & resources:
SIFo Medical YouTube Channel: Short, valuable videos on Quality Management
MedTech Free Resources: Get free access to checklists & templates
TMV Guide: Your practical guide to perform test method validation (incl. templates & videos)
References
[1] Taylor, Wayne (2017). Statistical Procedures for the Medical Device Industry. Taylor Enterprises, Inc., www.variation.com
[2] http://www.imdrf.org/docs/ghtf/final/sg3/technical-docs/ghtf-sg3-n99-10-2004-qms-process-guidance-04010.pdf
[3] http://www.quality-on-site.com/get.php?fileid=139
[4] ISO 11607-2:2019 Packaging for terminally sterilized medical devices — Part 2: Validation requirements for forming, sealing and assembly processes
[5] ISO 13485:2016 Medical devices — Quality management systems — Requirements for regulatory purposes
[6] https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart=820&showFR=1