Moving from prototype to mass production in MedTech rarely fails because of insufficient testing. Much more often, the real issue is the lack of a strategy capable of distinguishing necessary tests from tests that only create an illusion of safety.
At this stage, a medical device stops being merely an engineering project and becomes a system responsible for repeatability, safety, and, in many cases, therapeutic effectiveness.
In its simplest definition, validation answers one question:
In practice, however, especially in the MedTech industry, this question proves too simplistic. A far more accurate one would be:
Do we understand the process well enough to predict its behavior under conditions of variability that cannot be entirely eliminated?
At the level of individual components, the answer is usually yes.
Processes such as plastic injection molding are now well understood, and engineers have access to tools that allow them to control them effectively. Defining the process window, analyzing the influence of parameters on part quality, or assessing process capability are standard engineering practices that enable a high level of predictability.
Within this environment, the relationship between cause and effect remains relatively clear. A parameter change leads to a specific change in the molded part, while the process itself remains stable under proper supervision.
In practice, injection molding process validation is not limited to confirming that a part meets requirements under nominal conditions. A critical element is demonstrating that the process remains stable even under controlled parameter deviations within the Min / Nom / Max range.
This approach has not only a quality dimension, but also a business one. A properly defined process window ensures that minor changes resulting from tool wear, material variation, or environmental conditions do not immediately lead to a loss of process control.
In other words, validation creates an operational buffer that allows production to continue without immediate intervention whenever slight deviations from nominal values occur.
And it is precisely this experience of process stability at the component level that often creates the misleading assumption that the same approach can be directly transferred to the final product level.
This is where the natural belief emerges that if every component meets its specifications, then the entire product should function correctly as well.
And this is exactly where the problem begins.
Once the process moves to the assembly level, this assumption is no longer true.
The final product is not a simple sum of components, but rather the result of their interactions. Variability that remains acceptable at the level of a single component begins to accumulate and reveal itself in ways that cannot be predicted by analyzing individual parts separately.
In practice, this leads to situations where all components pass quality inspection, yet the final product still fails to meet functional requirements.
From the perspective of the project team, this becomes a particularly difficult moment, because traditional engineering tools are no longer sufficient and the sense of process control starts to disappear.
A good example of this complexity can be found in DPI (Dry Powder Inhaler) devices.
These medical devices rely heavily on the repeatability of the entire mechanical system. Even minor deviations may influence device performance and therefore its functionality.
Such devices often consist of more than 20 components that must interact predictably under various operating conditions.
In this context, validation is no longer only a quality issue — it becomes directly linked to product safety and effectiveness. This approach is also fully aligned with risk-based methodologies required by both ISO 13485 and ISO 14971.
What is crucial, however, is that the assembly process does not operate independently from earlier manufacturing stages.
Every component introduced into assembly carries variability originating from the injection molding process, even if it remains within the validated process window.
This means that assembly validation does not start from zero — it builds upon already existing variation present in individual components.
And it is precisely the accumulation of these deviations that begins to play a critical role at the level of the complete product.
In this context, validation is not merely testing the final product. It is confirming that the entire manufacturing process is capable of repeatedly delivering the expected results.
In other words, the assembly process does not eliminate variability — it aggregates it.
If every component may exist in several acceptable variants, the number of possible configurations grows exponentially.
Let us consider a simple example:
23 components,
each existing in three variants: minimum, nominal, and maximum.
The number of possible combinations equals:
3²³ = 94,143,178,827
More than 94 billion possible configurations.
Even assuming one test per minute, this would require more than 170,000 years of testing.
At this point, project teams often react naturally: “Let’s test everything.”
And this is exactly where mathematics stops being a tool and starts exposing the absence of a validation strategy.
In practice, this leads to months of delays, increasing testing costs, and loss of control over the project timeline.
At this stage, one critical question appears: Do all combinations truly matter equally?
The answer is no.
And this is where the difference begins between validation based on the number of tests and validation based on understanding the system.
In complex medical devices, only selected components and only certain interactions between them have a real impact on final product functionality.
These are the interactions that determine factors such as system sealing, repeatability of mechanical operation, positioning accuracy, or assembly forces.
Other components, although necessary, do not generate the same level of risk.
The problem is that without a conscious analysis, all components tend to be treated equally.
As a result, validation stops being a tool for risk reduction and instead becomes an attempt to compensate for uncertainty through the sheer number of tests performed.
A risk-based approach changes the perspective entirely.
Instead of asking: “What else can we test?” the question becomes: “What can go wrong, and what would be the consequences?”
In practice, this means moving away from analyzing isolated components toward understanding the entire system and its critical interfaces.
The key becomes identifying those areas where accumulated tolerances and variations may lead to a loss of functionality.
At this point, validation stops being a collection of tests and becomes a consciously designed decision-making process.
One of the tools commonly used in this approach is the so-called Matching Min / Nom / Max methodology.
When applied correctly, it does not involve mechanically testing every possible combination. Instead, it focuses on consciously selecting scenarios that represent the most unfavorable operating conditions of the system.
In other words, the objective is not to prove that everything works under ideal conditions, but to verify whether the system still performs correctly in its most demanding configurations.
This is the fundamental difference that determines the quality of the entire validation strategy.
Instead of billions of meaningless tests, the process is reduced to several or several dozen carefully selected cases that genuinely provide engineering value.
And this is precisely the moment when validation based on mathematics becomes validation based on engineering.
Validation is not about the number of tests performed. It is about the quality of the decisions behind them.
In regulated environments, the greatest risk is not testing too little.
The greatest risk is losing the understanding of what is actually being tested.
Without process understanding, there is neither quality nor safety — only an illusion of control that lasts until the system stops behaving predictably.
Krzysztof Osowski
Board Representative for MedTech Development at BIAZET S.A.
Expert in process validation, industrialization, and contract manufacturing
