Independent Batch Testing: Sampling, Chain of Custody and Reproducibility

Independent testing reduces reliance on a supplier’s paperwork, but independence is only one part of a defensible evidence chain. The result must also be connected to representative sampling, documented custody, a fit-for-purpose method and transparent reporting.

Every batch or periodic verification? Separate testing frequency, representative sampling and analytical coverage before interpreting a batch claim.

The evidence chain

1. Define the lot: identify the manufacturing or packaging population the conclusion is intended to cover.
2. Select the sample: use a documented plan that addresses representativeness and selection bias.
3. Seal and identify: apply unique identifiers and tamper-evident controls.
4. Transfer custody: record who handled the sample, when, why and under what conditions.
5. Test appropriately: use a laboratory and method competent for the exact analytical question.
6. Report transparently: link raw data, calculations, results and limitations to the same sample.
7. Interpret narrowly: do not generalise beyond what the sampling plan and method support.

Why supplier-selected samples are limited

If a supplier chooses and submits the unit, an external laboratory can still produce a technically independent result. However, the laboratory usually cannot establish that the selected unit was random or representative of every item in the lot.

The conclusion should therefore say “the submitted sample met…” rather than imply that the laboratory audited the entire production batch.

Every batch or periodic testing: what does risk-based sampling mean?

Testing frequency, sample selection and analytical coverage are three separate decisions. Testing every batch does not mean testing every unit. Periodic verification does not make an untested batch equivalent to a tested one. A defensible conclusion identifies which decision the evidence actually supports.

The WHO sampling guideline, sections 1.1–1.3, ties sampling to its purpose and distinguishes a representative sample from a deliberately selected sample. An independent laboratory can analyse a submitted unit competently while having no evidence that the sender selected it representatively. Repeated measurements of that same unit cannot supply the missing lot coverage.

What does periodic or skip testing actually allow?

ICH Q6A, section 2.1, describes a reduced frequency for specified tests on preselected batches or intervals. Untested batches must still meet the applicable acceptance criteria. In its drug-approval setting, this approach needs justification and prior regulatory approval, is generally introduced after approval, and can require a return to batch-by-batch testing when findings warrant it.

This is not permission to omit every test after one satisfactory certificate. Q6A concerns specifications for new chemical drug substances and products; it does not cover clinical-research stages and is insufficient for higher-molecular-weight peptides and biological products. Its scope matters before applying the concept to any peptide material.

What evidence belongs in a risk-based justification?

ICH Q9(R1) bases quality risk management on scientific knowledge and the protection of patients, with effort and documentation proportionate to risk. Uncertainty and the consequences of a wrong decision matter. The TGA records adoption of Q9(R1) on 24 May 2026; that adoption does not certify a particular material or sampling plan.

These questions are an editorial reading framework, not a prescribed sample size, acceptance standard or release procedure. A qualified quality unit must justify the actual plan for its material, purpose and regulatory context.

A worked claim check: one sample is not a whole batch

Illustrative scenario, not experimental data: a sender chooses one unit from lot A and an external laboratory reports an HPLC area percentage. The result describes the submitted sample under the stated method. Without documented selection, it does not establish that the unit represents all of lot A, cover lot B, or supply the separate evidence needed for identity, absolute content or microbiological quality. See our HPLC interpretation guide for the measurement limits.

If an unexpected result appears, changing the sample until one passes is not a scientific justification. The FDA’s May 2022 out-of-specification guidance, section IV.B, distinguishes retesting from resampling and calls for a documented basis for additional testing. A later passing result does not by itself invalidate an earlier valid failure.

Sources checked for this sampling clarification: 26 October 2026. The cited documents have their own publication and adoption dates; this is not a new testing standard or a product certification.

What a sampling plan should define

• the population or lot being sampled;
• how units are selected and by whom;
• the number of increments or units;
• whether a composite or individual units are tested;
• acceptance criteria and actions after an out-of-specification result;
• retention samples and conditions;
• controls against substitution or tampering.

Sample size should reflect the decision being made, expected variability and relevant standard—not a universal number copied from another context.

Chain of custody

A chain-of-custody record connects the sampled unit to the result. It commonly includes:

• unique sample and seal identifiers;
• lot, container and location information;
• collection date, time and sampler;
• condition and photographs where appropriate;
• every transfer, recipient and purpose;
• transport temperature or environmental controls;
• receipt condition at the laboratory;
• storage, sub-sampling and disposal records.

A gap does not necessarily invalidate all data, but it weakens claims about provenance and should be disclosed.

Independence, competence and accreditation

Organisational independence reduces some conflicts of interest. Technical competence determines whether the result is reliable. Accreditation can provide third-party assurance for listed activities, but only within the laboratory’s published scope.

For Australian laboratories, use the NATA directory to verify the organisation and scope. A logo alone is insufficient.

Reproducibility and measurement uncertainty

A measurement has variability. Precision within one run, between runs, between analysts and between laboratories can differ. A result close to a specification limit should be interpreted with method precision and, where relevant, measurement uncertainty in mind.

Independent repeat testing is most informative when the second laboratory receives a separately controlled sample and uses an appropriate method. Identical numbers are not expected; scientifically compatible results are the goal.

Out-of-specification results

An unexpected result should trigger a documented investigation, not automatic retesting until a passing number appears. The review may examine calculation, instrument performance, standards, sample preparation, method execution and sample integrity. Any invalidation needs an assignable, documented scientific reason.

If additional testing is performed, all valid results and the rationale for the testing plan should remain visible.

How to communicate a batch claim

A responsible summary names:

• the exact lot and sample tested;
• who selected and submitted it;
• the laboratory and scope status;
• the method and numerical result;
• the predefined criterion;
• the sampling and custody limitations;
• the report date and version.

Avoid marketing phrases such as “verified safe,” “pharmaceutical grade” or “guaranteed pure” unless every part of that claim is defined and supported by the appropriate evidence.

Related guides

• How to Read a Peptide Certificate of Analysis
• What HPLC and Mass Spectrometry Can—and Cannot—Prove
• Laboratory Research Tools

Technical reference

• ICH Q2(R2): Validation of Analytical Procedures
• NATA: Find an accredited organisation

Editorial note: This article explains evidence quality. It does not validate a particular batch, laboratory or supplier.