1What the regulations say about human error.
EU GMP names human error directly. Chapter 1 says that “where human error is suspected or identified as the cause, this should be justified having taken care to ensure that process, procedural or system-based errors or problems have not been overlooked, if present”.¹ Chapter 8 repeats the point for quality defects, and asks for the finding to be “formally justified”.² Both chapters ask for “an appropriate level of root cause analysis” and, where the true root cause cannot be determined, for the most likely root cause to be identified and addressed.¹²
The US regulation does not mention human error, but it sets the standard any finding has to meet. Under 21 CFR 211.192, any unexplained discrepancy, or a batch failing a specification, “shall be thoroughly investigated”, the investigation extends to other batches and products that may be associated with the failure, and the written record includes “the conclusions and followup”.³
ICH Q10 asks for “a structured approach to the investigation process” with “the objective of determining the root cause”, with effort and formality commensurate with the level of risk, and says CAPA should result in product and process improvements.⁴ WHO’s GMP text likewise asks for an appropriate level of root cause analysis and for the most likely root cause to be identified.⁵
So human error can be a finding, but not a conclusion an investigation reaches by default. It has to be shown, and the investigation has to show it looked for the process, procedure or system behind the error.
2What FDA has cited.
A 2025 FDA warning letter shows what an unjustified finding looks like. In warning letter 320-25-88 of 2 July 2025, citing 21 CFR 211.192, FDA wrote that a firm’s investigation “documented human performance as the cause of the failing testing events … however, no specific human error was identified”. In another investigation at the same firm, “although you identified the root cause as related to human error, no corrective action or training was implemented to prevent recurrence.” Elsewhere in the letter, FDA said the firm’s response did not provide evidence of “how you determined human error was the sole reason for the failed tests”.⁶
FDA’s own device training makes a similar point about the corrective action. In CDRH Learn’s module on CAPA, the FDA presenter says that when, as an auditor, he saw “training being used over and over and over again as a corrective action”, he would question how much investigation had been done to determine the cause, and at least question the training programme as a whole.⁷
3What a justified finding shows.
An investigation that concludes human error should be able to answer each of these from evidence.
| Question | What the regulation or guideline asks |
|---|---|
| Was the instruction clear? | Instructions and procedures are written “in an instructional form in clear and unambiguous language, specifically applicable to the facilities provided”, laid out so they are easy to check.¹⁸ |
| Was the person trained for this task? | Training covers the particular operations the employee performs and the written procedures that apply, on a continuing basis.⁹ Its practical effectiveness is periodically assessed.¹⁰ |
| Did the process or system make the error possible? | Process, procedural or system-based errors have been looked for and are not overlooked.¹² |
| What else could it have affected? | The investigation extends to other batches and products that may be associated with the failure.³² |
| Has it happened before? | Previous reports are reviewed for specific or recurring problems, and defect records are trended regularly.² |
| What will stop it recurring, and did that work? | Appropriate CAPAs are identified and taken, and their effectiveness is monitored and assessed.¹² |
A finding that names the specific error, explains why it was possible and links to an action that addresses that reason meets these questions. A finding that stops at “operator error” does not.
4Why retraining alone rarely closes it.
Retraining treats the person, while EU GMP asks the investigation to rule out the process, the procedure and the system.¹ ICH Q10 expects CAPA to result in product and process improvements and enhanced understanding, and in commercial manufacturing it asks for the effectiveness of the actions to be evaluated.⁴ If the same error recurs after retraining, a gap in training was not the whole cause, and an effectiveness review should show that.
Where the true cause cannot be found, EU GMP does not ask an investigation to invent one. It asks for the most likely root cause or causes to be identified and addressed, which keeps the uncertainty on the record instead of filling it with a label.¹
5How it runs in Seal.
In Seal’s deviation workflow, where a human contribution is suspected, the investigation plan assesses task design, procedure, interface, tools, workload and other relevant conditions, and records the evidence for and against; a “human error” label alone does not explain the mechanism. An investigation can end without a root cause, keeping the tested hypotheses, the missing evidence and the residual risk rather than forcing one.
Neil, Seal’s AI agent, can draft the investigation as a sourced chronology with the open questions kept visible; the investigator determines root cause and QA approves the investigation and any CAPA. Where the action is training, Seal’s training records can require observed practice or a knowledge assessment rather than a reading acknowledgement, tied to the procedure version.
References
- 1EudraLex Volume 4, Chapter 1: Pharmaceutical Quality System, sections 1.4(xiv) and 1.8(iv), European Commission, in operation since 31 January 2013. https:/
/ health. ec. europa. eu/ system/ files/ 2016- 11/ vol4- chap1_ 2013- 01_ en_ 0. pdf (read 6 November 2026) - 2EudraLex Volume 4, Chapter 8: Complaints, Quality Defects and Product Recalls, sections 8.11 to 8.19, European Commission, in operation since 1 March 2015. https:/
/ health. ec. europa. eu/ document/ download/ b1eb2292- cb0d- 4e3f- aea9- e3fe79faf6e3_ en?filename=2014- 08_ gmp_ chap8. pdf (read 6 November 2026) - 321 CFR 211.192, Production record review, eCFR. https:/
/ www. ecfr. gov/ current/ title- 21/ chapter- I/ subchapter- C/ part- 211/ subpart- J/ section- 211. 192 (read 6 November 2026) - 4ICH Q10, Pharmaceutical Quality System, section 3.2.2, ICH, 4 June 2008. https:/
/ database. ich. org/ sites/ default/ files/ Q10%20Guideline. pdf (read 6 November 2026) - 5WHO good manufacturing practices for pharmaceutical products: main principles, WHO Technical Report Series No. 986, Annex 2, World Health Organization, 2014. https:/
/ cdn. who. int/ media/ docs/ default- source/ medicines/ norms- and- standards/ guidelines/ inspections/ trs986annex2. pdf (read 6 November 2026) - 6Warning Letter 320-25-88, Exela Pharma Sciences, LLC, FDA Center for Drug Evaluation and Research, 2 July 2025. https:/
/ www. fda. gov/ inspections- compliance- enforcement- and- criminal- investigations/ warning- letters/ exela- pharma- sciences- llc- 709859- 07022025 (read 6 November 2026) - 7Corrective and Preventive Actions, Transcript, CDRH Learn, FDA, content current as of 4 January 2018. https:/
/ www. fda. gov/ training- and- continuing- education/ cdrh- learn/ corrective- and- preventive- actions- transcript (read 6 November 2026) - 8EudraLex Volume 4, Chapter 4: Documentation, section 4.4, European Commission, January 2011. https:/
/ health. ec. europa. eu/ system/ files/ 2016- 11/ chapter4_ 01- 2011_ en_ 0. pdf (read 6 November 2026) - 921 CFR 211.25, Personnel qualifications, eCFR. https:/
/ www. ecfr. gov/ current/ title- 21/ chapter- I/ subchapter- C/ part- 211/ subpart- B/ section- 211. 25 (read 6 November 2026) - 10EudraLex Volume 4, Chapter 2: Personnel, section 2.11, European Commission, in operation since 16 February 2014. https:/
/ health. ec. europa. eu/ document/ download/ 11f4f8e6- a6e9- 4897- afe3- f21e1dc56cb8_ en?filename=2014- 03_ chapter_ 2. pdf (read 6 November 2026)
AQuestions and answers
Can human error ever be the root cause of a deviation?
Yes, if it is justified. EU GMP accepts human error as a finding when the investigation has made sure that process, procedural or system-based errors have not been overlooked. The record should name the specific error, show the evidence for it and explain why other causes were excluded.
Is retraining an acceptable CAPA for human error?
It can be part of one, but on its own it rarely addresses why the error was possible. An FDA trainer has said that seeing training used again and again as a corrective action makes him question the investigation. Check whether the error recurs after the action, and change the procedure, form or system if that is where the cause lies.
What is the difference between a human error and its root cause?
The error is what the person did, such as skipping a step or entering the wrong value. The root cause is why that was possible or likely, for example an ambiguous instruction, a form that did not prompt for the check, or a step the system did not verify. Regulators expect the investigation to look for that cause rather than stop at the error.
What if the investigation cannot find the true root cause?
EU GMP asks you to identify the most likely root cause or causes and to address those. Record what was tested, what evidence was missing and how confident the conclusion is, so the uncertainty stays visible and the effectiveness review can test it.
Should human-error findings be trended?
Yes. EU GMP asks for quality defect records to be reviewed and trended regularly for recurring problems, and ICH Q10 includes deviations in process and product monitoring and in management review. Trending human-error findings by task, procedure or step shows where the system makes errors likely.