The Science of Sampling — Statistical Rigor for CCI
Sampling is where statistical theory meets operational reality. For container closure integrity, the sampling plan is the foundation of every release decision — and when built thoughtfully, it provides a statistically robust, operationally feasible, and regulatorily defensible path to quality assurance.
The recurring question is simple: How many units must be tested to be confident in the quality of the batch? The answer requires honest engagement with three competing pressures. Confidence interval sampling based on binomial probability can produce a mathematically defensible number — but a 95 percent confidence interval at a 0.01 percent defect rate requires nearly 30,000 samples. For destructive CCI methods, that number is operationally impossible. It would destroy a commercial lot.
Squeglia's Zero Acceptance Number Sampling Plan (c=0) resolves that tension. Grounded in ANSI/ASQ Z1.4 and widely accepted by global regulators, the c=0 plan is prospective: the sample size is designed up front to control the probability of accepting a non-conforming lot. The rule is simple — test n units, reject the lot if any defect is found. The math yields approximately 1,250 samples to provide 95 percent confidence that a defect rate of 1 in 10,000 will not escape detection. Operationally sustainable. Statistically defensible. Regulatorily accepted.
The difference between Squeglia's approach and a post-hoc confidence interval matters. Post-hoc intervals ask, "Based on these results, what is the likely true defect rate?" That framing is inherently conservative because it does not control sampling prospectively. Squeglia's plan controls acceptance risk through the design of the plan itself — a distinction that aligns precisely with modern Quality Risk Management expectations under ICH Q9, USP <1207>, and EMA Annex 1.
The Operating Characteristic (OC) curve is the visual defense of the plan. It shows probability of lot acceptance against true defect rate. A well-designed c=0 plan produces a curve that drops steeply near the Acceptable Quality Limit, demonstrating that lots with unacceptable defect rates will be rejected with high confidence. The OC curve is the answer to every auditor question about sample size justification.
The final discipline is recognizing that a sampling plan is only as strong as the test method powering it. A high signal-to-noise ratio method with a smaller sample can outperform a lower-quality method applied to more units. The combination of Squeglia's c=0 plan with deterministic, quantitative test methods forms the defensible, efficient, patient-centered sampling strategy the modern regulatory landscape demands.
Frequently Asked Questions
1. Why is Squeglia’s c=0 sampling plan widely used for CCI?
Squeglia’s zero acceptance (c=0) plan provides a practical balance between statistical rigor and operational feasibility. Instead of requiring impractically large sample sizes, it prospectively controls acceptance risk—rejecting a lot if any defect is found. This approach aligns with standards like ANSI/ASQ Z1.4 and regulatory expectations for risk-based decision-making.
2. 2: How is Squeglia’s approach different from confidence interval sampling?/strong>
Confidence interval sampling is retrospective—it interprets results after testing and often leads to very large, impractical sample sizes. In contrast, Squeglia’s c=0 plan is designed upfront to control the probability of accepting defective lots. This prospective design aligns with modern frameworks such as ICH Q9 and USP <1207>.
What role does the test method play in sampling effectiveness?
The reliability of a sampling plan depends heavily on the sensitivity and consistency of the CCI method used. Deterministic, high signal-to-noise methods can detect smaller defects with fewer samples, making the overall strategy more efficient and defensible—especially under evolving regulatory expectations like EU GMP Annex 1.