CPVContinued Process Verification

Continued Process Verification is the third stage of FDA's 2011 Process Validation guidance — the ongoing programme that monitors the commercial process throughout the product's lifecycle, ensures it stays in the validated state, and feeds the Annual Product Quality Review (APR) under 21 CFR 211.180(e). It is not a one-time activity; it is the operating mode for every batch after PPQ closes.

EU GMP Annex 15 calls the same concept Ongoing Process Verification (OPV) and ICH Q10 §3.2.5 calls it 'process performance and product quality monitoring system'. The terminology differs; the practice is identical.

• Critical process parameters (CPPs) and critical quality attributes (CQAs) monitored, with the rationale (typically a flow-down from the Stage 1 control strategy).
• Sampling plan — initial elevated sampling (typically first 10–30 commercial batches), step-down criteria, routine sampling.
• Control limits — typically ±3σ from PPQ mean, narrower than release spec, used as out-of-trend (OOT) signals.
• Statistical tools — control charts (X-bar/R, Individuals & Moving Range, EWMA), capability indices (Ppk on rolling windows), trend tests (Western Electric rules, Nelson rules).
• Action plan for each signal type — OOT investigation, CAPA, change-control trigger, when to escalate to a process revalidation.
• Review cadence — monthly trend review at minimum, quarterly at management review, annual at APR.

Where PPQ reports short-term Cpk on the qualification batches, CPV reports long-term Ppk on a rolling 12-month (or quarterly) window. The shift from σ_within to σ_overall captures long-term variability — between-shift, between-operator, between-supplier-lot, seasonal — that PPQ deliberately controlled out. Ppk is therefore always ≤ Cpk; the gap is the long-term penalty.

Practically: every commercial batch's CQA results land in the same dataset, the rolling Ppk recomputes monthly, and the trend chart shows whether the process is stable, drifting, or improving. A Ppk that drifts from 1.50 to 1.20 over six months is a CAPA trigger long before any individual batch fails release.

Single-point excursion past a control limit is the obvious signal, but most process drift shows up earlier as a pattern within the limits. The Western Electric Rules (1956) and the expanded Nelson Rules (1984) define eight patterns that indicate a process is no longer in statistical control even when no single point is OOS:

1. One point beyond 3σ — special-cause variation.
2. Nine consecutive points on one side of the centreline — process mean has shifted.
3. Six consecutive points trending up or down — directional drift.
4. Fourteen consecutive points alternating up and down — non-randomness.
5. Two of three consecutive points beyond 2σ on the same side.
6. Four of five consecutive points beyond 1σ on the same side.
7. Fifteen consecutive points within 1σ — over-control or measurement saturation.
8. Eight consecutive points beyond 1σ in either direction — overdispersion or mixed populations.

Modern CPV systems apply these rules automatically across thousands of attribute streams; manual chart review on every product every month is impossible at scale.

21 CFR 211.180(e) and EU GMP Chapter 1 §1.4(xv) require an annual review of every commercial product covering: batches produced, releases and rejections, complaints and recalls, deviations and CAPAs, stability data, change controls, OOS and OOT events, and a summary of process performance. The APR is not a separate exercise from CPV — it is the annual consolidation of the same data the CPV programme tracks monthly.

A well-designed CPV programme makes the APR a 30-day exercise instead of a 90-day spreadsheet ordeal. A poorly designed one means quality teams spend Q1 of every year scraping data out of disconnected systems.

1. CPV protocol not written — Stage 3 monitoring is ad hoc.
2. Control limits set at the release spec — no signal until OOS, too late.
3. Trend rules not applied — single-point OOS is the only flagged event.
4. Rolling Ppk not computed — capability degradation invisible.
5. OOT investigations closed without root cause — pattern repeats.
6. Sampling step-down without statistical justification — reduced sampling has no defensible basis.
7. APR assembled from Excel exports rather than live CPV data — risk of transcription error and stale data.
8. Process change implemented without updating CPV control limits — old limits applied to new process produce false signals.

• Every CQA from every commercial batch lands in the CPV dataset by construction — no manual export.
• Control limits flow from the PPQ closure into the CPV charts; a process change creates a new limit set under change control.
• Western Electric and Nelson rules run on every chart; OOT signals open investigation tickets with the rule and the chart attached.
• Rolling Ppk per CQA per product computes monthly; drift triggers a CAPA before an OOS.
• Sampling step-down rules are configurable per product; the step-down event is logged with the statistical justification.
• The Annual Product Quality Review template pulls every section (batches, complaints, deviations, OOS, OOT, change controls, stability, capability) from the same database — APR drafting is review-and-sign, not assemble-from-scratch.