The QC Lab

A specification is the written list of tests, methods and acceptance limits for a material or product: assay 95.0–105.0% of label claim, pH 5.5–7.0, total aerobic count ≤ 100 CFU/g, and so on. The chemist never decides 'is this good?' from opinion — the spec decides; the chemist produces trustworthy numbers against it. Specs come from pharmacopeias (USP monographs), FDA-approved filings for drugs, and validated internal methods for cosmetics and supplements.

If it isn't in the spec, it isn't a release requirement. If it is in the spec, there are no exceptions without a formal investigation.

Each incoming lot arrives with the supplier's CoA: their results for that lot. QC's job is to decide how much of it to re-verify in-house. Critical identity testing is almost always done in-house regardless of the CoA; full re-testing frequency is set by the supplier's qualification history — a proven supplier may earn reduced (skip-lot) testing, a new one gets full panel testing on every lot.

A CoA with results suspiciously identical lot after lot is itself a red flag a good chemist notices.

Testing is destructive and slow, so QC tests samples chosen to represent the lot. Raw-material containers are commonly sampled on a √N+1 plan; finished goods use ANSI/ASQ Z1.4 sampling with an Acceptance Quality Limit (AQL) — e.g., AQL 0.65 for critical defects means even a small count of critical defects in the sample rejects the lot. The math guarantees that bad lots are very unlikely to slip through while good lots aren't rejected by bad luck.

Sampling technique matters as much as sample size: thief samplers for powders, aseptic technique for micro samples, and a documented chain of custody.

An OOS result starts a regulated investigation, in two phases. Phase I is the lab: was it the analyst, the instrument, the standard, the calculation? Only a proven assignable lab cause can invalidate a result. Phase II goes to production: was it the batch itself? Retesting follows a pre-approved protocol — 'testing into compliance' (rerun until it passes) is explicitly prohibited and is one of the most-cited FDA violations in the industry.

The chemist's data integrity — original results are never deleted or overwritten — is what makes the whole system legally defensible.

Every number in the lab must be Attributable, Legible, Contemporaneous, Original and Accurate (ALCOA), plus Complete, Consistent, Enduring and Available. In practice: signed and dated entries made at the bench at the time of the work, instrument audit trails on, second-person verification of calculations, and no scrap-paper 'rough notes' that later get recopied. Most modern QC labs run chromatography data systems (like Empower) that enforce this electronically.

When people picture 'quality,' this paperwork discipline — not the beakers — is most of what inspectors actually audit.

When something fails, a Corrective and Preventive Action (CAPA) file documents the root cause and the fix so the same failure can't recur silently. And nothing in a validated process changes casually: a new supplier, a new instrument, a tweaked method all go through change control, with re-validation proportional to risk. This is how a QC system learns instead of just catching.

Trending is part of it too: results that all pass but drift steadily in one direction (out-of-trend, OOT) get investigated before they become failures.