Process validation and equipment qualification for MedTech suppliers: DQ, IQ, OQ and PQ in practice
If you make parts or assemblies for medical devices, sooner or later a customer asks: is this process validated? Cleaning, passivation, laser marking, sealing sterile packaging or bonding often cannot be fully checked on the finished part. In that case the process itself has to show that it reliably produces good parts.
This article explains when process validation is needed, what DQ, IQ, OQ and PQ each prove, and why a PQ without several operators can give a false picture. The example at the end comes from a real qualification project.
Summary
- A process is validated when its result cannot be fully verified by subsequent inspection (ISO 13485, clause 7.5.6).
- DQ proves that the design meets the requirements, IQ the correct installation, OQ the behaviour at the process limits, PQ the repeatability under real production conditions.
- Real conditions include different days, material lots and operators. Operators are often the largest variable.
- Sample size and acceptance criteria are derived from risk and justified before execution.
When a process has to be validated
ISO 13485 clause 7.5.6 requires validation of processes whose output cannot be or is not verified by subsequent monitoring or measurement. Typical cases are processes where full inspection would destroy the part, or where defects only become visible in use. Manufacturers under the EU MDR base their quality management system on ISO 13485 and pass this requirement on to their suppliers, often through quality agreements and supplier audits.
The question is therefore not only whether your process produces good parts today, but whether you can show that it does so consistently under all expected conditions.
Typical supplier processes
- Cleaning of instruments and components before packaging or sterilization
- Passivation and other surface treatments
- Laser marking (legibility, UDI, effect on the surface)
- Sealing of sterile barrier systems (validation to ISO 11607-2)
- Bonding, welding, soldering and crimping
- Injection moulding and other forming processes with critical features
DQ, IQ, OQ and PQ: what each stage proves
DQ (design qualification) shows that the design of the equipment meets the requirements, before it is built or ordered. The basis is the user requirements specification (URS) with product and process requirements, cleanroom and safety requirements. The equipment supplier's specification, drawings and functional description are checked against it point by point, ideally in a traceability matrix, and risks from the risk analysis are addressed in the design. Whatever is missed in the DQ otherwise only shows up in the OQ or PQ, when changes are expensive.
IQ (installation qualification) shows that the equipment is installed to specification: utilities, software and firmware versions, calibrated measuring equipment, documentation and safety functions. IQ freezes the baseline that all later evidence refers to.
OQ (operational qualification) shows that the process produces good results across the whole process window, including the upper and lower parameter limits (worst case). This is where the allowed ranges for temperature, pressure, time or speed are set.
PQ (performance qualification) shows that the process repeatedly meets the requirements under real production conditions: with production material, over several days or lots, and with the people who will run production.
PQ in practice: an example from a qualification project
In a project to qualify two new cleanroom production machines that make a component of a catheter, we carried out the full qualification from DQ to PQ. In the PQ we tested all product combinations, with a sample of 10 parts per configuration. The details are under a non-disclosure agreement, so the description stays general. The results varied from day to day: on one day 10 out of 10 parts were good, on another day only 5 out of 10, with the same configuration and the same parameters.
The root cause analysis showed that neither the machine nor the parameters were the problem. The cause was the feeding: the way operators loaded the parts into the machine differed from person to person. Each operator had their own technique, and those small differences were enough to change the result.
The lesson is one that many validation plans miss: a PQ with a single operator, often the most experienced one, shows the performance of that operator, not of the process.
Operators belong in the PQ
Every operator has a bias: how a part is gripped, loaded, aligned or pressed. That is not a criticism, it is normal. A validated process has to cope with it. In practice this means:
- Plan PQ runs with several operators, ideally also across different shifts and days.
- Evaluate results per operator, not only as a total. A difference between operators is a finding, not chance.
- Describe manual steps such as feeding, loading and unloading in the work instruction precisely enough that two people do them the same way.
- Where possible, reduce the dependence on the operator technically, for example with a fixture, an end stop or a position check.
- Document operator training and qualification before the PQ starts.
The same principle applies to measurement: a GR&R study (measurement system analysis) separates process variation from variation caused by inspectors and gauges. More in the article GR&R and MSA for test stands.
Sample size and worst case
Ten parts per configuration are often a sensible first look and make problems like the one above visible quickly. For a statistical statement they are often not enough. The required sample size depends on the risk of the characteristic, the required reliability and confidence, and the type of data (attribute or variable).
A common approach for attribute characteristics is the success-run theorem: 95 % reliability at 95 % confidence requires 59 parts with no failures. With variable data, process capability indices (Cpk, Ppk) can often justify smaller samples. What matters is that sample size and acceptance criteria are defined and justified in the validation plan before execution, not afterwards.
In the PQ, worst case means not only parameter limits, but also the least favourable product variant, material lot or part geometry that occurs in production.
What auditors and customers ask
- Why does this process require validation, and where is that justified?
- How were the requirements checked against the equipment design in the DQ?
- How were the process window and the worst case defined?
- How are the sample size and acceptance criteria justified?
- Which operators, shifts and material lots were included in the PQ?
- How were deviations investigated and closed?
- When will the process be revalidated?
When to revalidate
- Changes to equipment, software, tooling or fixtures
- New materials, suppliers or product variants
- Relocation of the equipment or changes to utilities and environment
- Trends in process data, complaints or repeated deviations
- New operator teams, where manual steps influence the result
Checklist before the PQ
- DQ complete: every URS requirement traced to the design
- IQ complete, open items assessed
- OQ documented with process window and worst case
- Work instruction describes all manual steps, including feeding
- Operators trained, several operators and days in the plan
- Sample size and acceptance criteria justified in advance
- Measuring equipment calibrated, measurement system assessed with GR&R where needed
- Handling of deviations defined
Support
Techkrates plans and supports equipment qualification and process validation for MedTech manufacturers and suppliers, on-site in Switzerland and Germany. Get in touch or read more about equipment qualification (DQ/IQ/OQ/PQ).