Why investigations decide CAPA success
A CAPA can only be as good as its investigation. If the root cause is wrong, the actions treat a symptom and the problem comes back. That is exactly what FDA keeps finding. An analysis of FY2025 FDA inspection data found that 21 CFR 211.192, “Failure to thoroughly investigate unexplained discrepancies and failures,” was the second most cited drug observation, with 164 citations (GMP Insiders). Recent warning letters say it directly: “your investigation was not sufficient to adequately determine root cause(s)” (FDA, Glenmark, 2025), and investigation conclusions “lack the necessary rigor and scope to thoroughly identify root causes” (FDA, Viatris, 2024).
What regulators expect from an investigation
- Drugs: Unexplained discrepancies “shall be thoroughly investigated, whether or not the batch has already been distributed,” and “A written record of the investigation shall be made and shall include the conclusions and followup” (21 CFR 211.192).
- Pharmaceutical quality system: “A structured approach to the investigation process should be used with the objective of determining the root cause,” with effort “commensurate with the level of risk” (ICH Q10).
- Medical devices: Under the QMSR, effective February 2, 2026, ISO 13485’s corrective action requirements apply, including determining causes of nonconformities. FDA notes that “Clauses 8.2.2, 8.5.2, and 8.3.1 of ISO 13485 address investigations of complaints” (Federal Register).
- Quality unit oversight: “The quality unit has the responsibility to ensure that investigations are scientifically sound” (FDA, Apotex, 2025).
The GHTF guidance on CAPA, still widely used by device makers, sums up the goal: “Ensure that causes are identified, not the symptoms” (GHTF SG3 N18:2010).
Root cause analysis tools
ASQ defines root cause analysis as “a systematic process used to identify the underlying causes of problems or nonconformities” (ASQ).
| Tool | How it works | Best for | Watch out for |
|---|---|---|---|
| 5 Whys | Ask “why” repeatedly until a fixable cause appears | Simple, single-path problems | Stopping too early or following one path when several causes interact |
| Fishbone (Ishikawa) | “Identifies many possible causes for an effect or problem and sorts ideas into useful categories” (ASQ) such as people, methods, machines, materials, measurement and environment | Brainstorming with a team | Listing causes without testing them |
| FMEA | Scores failure modes by severity, occurrence and detection | Process or design problems; prioritizing actions | Becoming a scoring exercise detached from evidence |
| Fault tree analysis | Logic tree of events that combine to cause a failure | Complex technical failures | Effort; needs good system knowledge |
| Data analysis | Trending, stratification, comparison of good and bad lots | Problems with available data | Correlation mistaken for cause |
No tool proves a root cause on its own. The cause should be confirmed with evidence: data, testing, reproduction of the failure or elimination of alternatives.
The “human error” trap
Human error is the most common and least useful root cause. GEN reported that “human error has been attributed to approximately 50% of quality incidents” in biopharma, citing industry estimates (GEN). FDA watches for it: its Office of Quality Surveillance tracks “The proportion of root causes associated with human error” as a signal of quality system effectiveness (FDA OQS).
FDA has rejected human error conclusions when problems recurred. In one warning letter it noted the firm “attributed the root cause of non-contemporaneous documentation of microbiological plates to human error,” and called its earlier CAPAs “ultimately ineffective” (FDA, Medgel, 2023).
When people make mistakes, ask why the system allowed it: unclear procedures, poor design of the task or equipment, time pressure, inadequate training or missing checks. Those are fixable causes.
Running a CAPA investigation: step by step
- Define the problem precisely: what, where, when, how many, and what is different from normal.
- Contain affected product and assess risk to patients and users.
- Assess scope: other lots, products, lines, sites and time periods that may be affected.
- Gather evidence: records, data, samples, interviews and observation.
- Analyze causes with a suitable tool and test candidate causes against the evidence.
- Confirm the root cause and document why alternatives were ruled out.
- Define actions that remove the cause and prevent recurrence elsewhere.
- Set effectiveness criteria before implementation.
- Document conclusions and follow-up as 211.192 and ISO 13485 require.
Investigation report checklist
- Problem statement specific and measurable
- Immediate containment and risk assessment recorded
- Scope assessment beyond the triggering event
- Evidence listed and attached
- Root cause method named and applied
- Root cause confirmed by evidence; alternatives ruled out
- Human error, if cited, traced to system causes
- Actions mapped to each root cause
- Effectiveness criteria and timeframe defined
- Quality unit review and approval documented
Check whether the evidence supports the conclusion
The most common investigation weakness is a conclusion the evidence does not support. AskDona reviews investigation reports and CAPA records against your procedure, 21 CFR 211.192, ICH Q10 or ISO 13485, flags unsupported root causes, missing scope assessments and vague actions, and cites the passage behind every finding.
Run a sample investigation report assessment on one closed investigation. AskDona shows the evidence; your quality team decides.
FAQ
What is root cause analysis in CAPA? A structured process to find the underlying cause of a problem so corrective action can prevent recurrence.
What are the most common root cause analysis tools? 5 Whys, fishbone diagrams and FMEA are among the tools ASQ lists (ASQ).
Is human error an acceptable root cause? Rarely on its own. Investigators should identify the system conditions that allowed the error.
What does FDA expect in an investigation? A thorough, documented investigation with conclusions and follow-up, scaled to risk (21 CFR 211.192).
How do I know I found the real root cause? Evidence confirms it, alternatives are ruled out, and the problem does not recur after corrective action.
References
All pages opened and checked by October 13, 2026.
| Source | Type | Date | Used for |
|---|---|---|---|
| GMP Insiders: FY2025 observations | Trade analysis | Jan 5, 2026 | 211.192 citations |
| FDA warning letter: Glenmark | Regulator | Jul 11, 2025 | Root cause |
| FDA warning letter: Viatris | Regulator | Dec 19, 2024 | Investigation rigor |
| eCFR: 21 CFR 211.192 | Regulation | Current | Investigation requirement |
| ICH Q10 (FDA) | Guideline | Apr 2009 | Structured approach |
| Federal Register: QMSR final rule | Regulator | Feb 2, 2024 | ISO 13485 clauses |
| FDA warning letter: Apotex | Regulator | Oct 31, 2025 | Quality unit |
| GHTF SG3 N18:2010 | Guidance | Nov 4, 2010 | Causes not symptoms |
| ASQ: RCA | Professional body | 2024 | Definition, tools |
| ASQ: RCA tools | Professional body | n.d. | Fishbone |
| GEN: Understanding deviations | Trade press | May 7, 2025 | Human error share |
| FDA OQS: PQS effectiveness | Regulator presentation | c. 2023 | Human error signal |
| FDA warning letter: Medgel | Regulator | Jul 20, 2023 | Human error rejected |