Root Cause Analysis

Root Cause Analysis (RCA)

Root cause analysis is the structured search for why a problem occurred, rather than what occurred. The failure mode is stopping too early. “Operator error” is almost never a root cause — it is the point at which the investigation gave up. The productive question is why the system permitted the error: was the instruction ambiguous, the training ineffective, the interface easy to misread, the workload unrealistic, the equipment poorly designed for the task?

Different techniques suit different problems. The 5 Whys works for simple linear causation but degenerates into guesswork if the answers are not evidence-backed. Fishbone (Ishikawa) diagrams organise candidate causes across categories — people, method, machine, material, measurement, environment — and are useful for structuring a team discussion. Fault tree analysis works backwards deductively from a defined failure. For complex events, none of these substitutes for reconstructing the sequence and testing hypotheses against evidence.

A defensible analysis has three properties: the cause is supported by evidence rather than opinion, it explains all the observed facts rather than some of them, and removing it would plausibly have prevented the event. If the proposed root cause fails any of those tests, the corrective action built on it will not hold — which is why weak RCA is the most common reason deviations recur after a CAPA was closed as effective.

• —“Operator error” is where an investigation stopped, not a root cause.
• —Techniques: 5 Whys (simple, linear), fishbone (structuring), fault tree (deductive from a defined failure).
• —Every step must be evidence-backed — unevidenced 5 Whys is structured guessing.
• —A valid root cause explains all the facts, not merely some of them.
• —Test it: would removing this cause plausibly have prevented the event?
• —Weak RCA is the usual reason a closed, “effective” CAPA fails to stop recurrence.

Expected under 21 CFR 211.192 (investigation of discrepancies), 21 CFR 820.100 / the FDA QMSR for devices, ICH Q10 §3.2.2 (CAPA), and EU GMP Chapter 1. The techniques themselves are quality-engineering methods, not regulatory instruments.