Process Improvement and Quality Tools
Domain 3 - Process Improvement and Quality Tools
This domain tests whether you can select the right tool, apply it correctly, and connect improvement to the quality system. Expect PDCA and DMAIC, the classic quality tools, FMEA, and root cause analysis in a GMP context.
Core rules
- PDCA underpins most frameworks. DMAIC (Define, Measure, Analyse, Improve, Control) is the Six Sigma structure; the Control phase is what prevents improvement from decaying - a frequent exam focus.
- Tool-to-purpose mapping: Pareto for prioritisation, check sheets for data collection, histograms for distribution, scatter plots for correlation, control charts for stability over time, fishbone and 5 Whys for cause hypotheses, flowcharts and SIPOC for process understanding, FMEA for prospective risk.
- Root cause analysis must reach a controllable systemic cause. "Operator error" is an observation, not a root cause, unless the analysis explains why the system allowed the error: procedure design, labelling, workload, training effectiveness, or absence of an interlock.
- Improvement in a GMP environment flows through change control. Any change to a validated process, method, or system needs impact assessment, approval, and where relevant revalidation.
- Continuous improvement is an ICH Q10 enabler, alongside knowledge management and quality risk management.
Worked example
A tablet line has 38 deviations in six months. A Pareto chart shows 71% arise from two steps: granulation endpoint and tablet weight control. A fishbone on granulation identifies six candidate causes; a 5 Whys on the largest ("endpoint judged visually") leads to "no in-process parameter defined because the process was transferred without development data". The improvement is not retraining but defining an endpoint parameter, running a small design-of-experiments study, and updating the control strategy through change control. FMEA then checks whether the new control introduces new failure modes.
Traps that cost marks
- Choosing FMEA when existing data would be better explained by a Pareto chart or control chart.
- Stopping at "human error" and proposing retraining as the only CAPA.
- Using a tool as a ritual - a fishbone with no verification of the candidate causes.
- Forgetting that improvement actions need change control and effectiveness checks.
- Reacting to common-cause variation as if it were special cause, and over-adjusting the process.