Risk Assessment for Computerised Systems: From GxP Impact to Testing Depth

Annex 11 says the extent of validation and data integrity controls should be based on a justified and documented risk assessment. In practice, that means a risk assessment that actually changes what you test.

Two levels of assessment

A useful approach separates system-level and function-level questions.

  • System level: does the system touch GxP data or decisions, and what is its potential effect on patient safety, product quality and data integrity? This sets the overall validation approach.
  • Function level: which specific functions or requirements carry the risk? This sets the depth of testing for each one.

Scoring and its limits

Many organisations use severity, probability and detectability to rank functions, for example through an FMEA-style table. Such scores are a means of structuring judgement, not a replacement for it. Subjective scoring is a recognised weakness, which is why ICH Q9(R1) emphasises reducing subjectivity and being clear about the basis for decisions.

Keep scales simple, define each level with examples relevant to your processes, and have more than one person from the process, IT and quality challenge the result.

Linking risk to the testing strategy

  • High risk: scripted testing with predefined expected results, including challenge of limits and error handling.
  • Medium risk: reduced scripting, with objective evidence retained.
  • Low risk: reliance on supplier testing or documented unscripted checks.

Annex 11 also expects test methods to cover parameter limits, data limits and error handling, so high-risk functions should show these challenges explicitly.

Keep it alive

A risk assessment completed once and filed is of limited value. Revisit it when requirements change, after incidents, and at periodic review. Include data integrity risks such as audit trail coverage, access control and data transfer, alongside functional ones.

Who should be in the room

Risk assessments written by one person at a desk tend to repeat the supplier’s marketing description. A short workshop with the process owner, an IT or system specialist and a quality representative usually surfaces the real failure modes: a calculation users do not check, an interface that fails silently, a report that feeds a release decision. Record the participants, the assumptions and the rationale for each rating, and keep the output linked to the requirements so the test strategy can be traced back to it.

How PQRA helps

PQRA can facilitate risk assessment workshops, design practical scoring methods consistent with ICH Q9(R1), and help you translate the outcome into a defensible, proportionate test strategy.

Contact PQRA to discuss how we can support your computerised system validation programme.

CATEGORIES:

Uncategorized

Tags:

No responses yet

Αφήστε μια απάντηση

Η ηλ. διεύθυνση σας δεν δημοσιεύεται. Τα υποχρεωτικά πεδία σημειώνονται με *