Skip-Lot and Periodic Testing: The Peptide COA Result That Was Never Run on Your Batch

Skip-lot testing is a formally recognised practice in which a specified test is performed only on pre-selected batches or at set intervals rather than on every batch, while every untested batch is still required to meet the same acceptance criteria. It means a line on a certificate of analysis can carry a number that was never generated from the vial in front of you. Under ICH Q6A the practice is legitimate, but only when it has been justified in advance and approved by a regulator, and only when the certificate itself makes clear which results came from the batch and which did not.

What Skip-Lot Testing Actually Is

The governing definition sits in ICH Q6A, section 2.1, titled Periodic or Skip Testing. The guideline defines it as the performance of specified tests at release on pre-selected batches and at predetermined intervals, rather than on a batch-to-batch basis, with the understanding that those batches not being tested still must meet all acceptance criteria established for that product. Q6A adds that this represents a less than full schedule of testing and should therefore be justified and presented to and approved by the regulatory authority prior to implementation.

Two features of that definition matter. The acceptance criteria never relax: a skipped attribute is still a specification requirement, and the manufacturer is asserting conformance on the basis of process knowledge rather than a fresh measurement. And approval is external. Skip testing is a regulatory dossier change, recorded in the Common Technical Document under section 3.2.S.4.1 for a drug substance or 3.2.P.5.1 for a finished product, and inspectors verify the approved frequency against what the site actually does.

The Pharmacopoeias Already Say Not Every Test Must Be Run

The European Pharmacopoeia General Statements, section 1.1, states that an article is not of Pharmacopoeia quality unless it complies with all requirements stated in the monograph, then immediately qualifies that this does not imply that performance of all tests in a monograph is necessarily a prerequisite for a manufacturer in assessing compliance with the Pharmacopoeia before releasing a product. The manufacturer may instead obtain assurance on the basis of the product design, the control strategy, and data derived from process validation studies.

The same position appears in the United States Pharmacopeia General Notices and in the WHO Expert Committee on Specifications for Pharmaceutical Preparations, thirty-sixth report, Annex 10 (WHO Technical Report Series No. 902, 2002): it is not to be inferred that application of every analytical procedure in the monograph to samples from every production batch is necessarily a prerequisite for assuring compliance with compendial standards before release. The thirty-seventh report (WHO Technical Report Series No. 908, 2003, page 87) restates it.

Key Research Findings

• ICH Q6A section 2.1 permits periodic or skip testing subject to prior regulatory justification and approval, and names residual solvents and microbiological testing as example candidate attributes.
• Swissmedic, presenting at its Workshop on Skip Testing (Bern, 12 to 13 June 2018), cited real industry proposals of one batch tested in every 20 to 30 batches, or one batch per year, whichever came first.
• Ph. Eur. General Statements 1.1 explicitly states that performing all monograph tests is not necessarily a prerequisite for a manufacturer to assess pharmacopoeial compliance before release.
• WHO Technical Report Series No. 902 (2002), Annex 10, and No. 908 (2003, page 87) contain the same position for compendial standards generally.
• The IPEC Certificate of Analysis Guide version 2.1 (2024), section 7.2, lists four permitted origins for a number printed on a certificate, only one of which is a test on a sample of the finished batch.
• The IPEC model certificate in Annex 1 marks a quarterly-tested attribute with nothing more than a footnote symbol, reporting arsenic as “NMT 2 ppm” against a specification of “NMT 2 ppm” rather than a measured value.
• ICH Q3C(R8) sets concentration limits of 410 ppm for acetonitrile (permitted daily exposure 4.1 mg), 600 ppm for dichloromethane (6.0 mg) and 880 ppm for N,N-dimethylformamide (8.8 mg), all solvents routinely present in solid-phase peptide synthesis.
• The statistical basis predates pharmaceutical use: Dodge, H. F. (1955), Industrial Quality Control 11(5), pages 3 to 5; SkSP-2 by Perry, R. L. (1970, Rutgers thesis); the attribute program by Liebesman and Saperstein (1983), Journal of Quality Technology 15(3), pages 130 to 139.

How a Compliant Certificate Signals a Skipped Test

The clearest published treatment of how this should appear on paper is the IPEC Certificate of Analysis Guide, version 2.1, issued in 2024. Written for pharmaceutical excipients rather than peptides, it nonetheless codifies reporting conventions adopted across the chemical supply chain.

The four permitted origins of a number on a certificate

Section 7.2 of the guide states that measurements reported on a certificate can be derived from testing a representative sample from the finished batch, from in-process testing of a representative sample where the attribute remains unaffected by further routine processing, from continuous monitoring of an attribute combined with statistical process controls, or from periodic monitoring of an attribute according to a defined plan. Where any of the last three apply, the guide says the technique for how the test result was obtained should be described.

Section 7.2 also permits a non-specific conformance statement in place of data when a specification test was not performed on the batch. The typical wording it lists includes “conforms”, the notably circular “if tested will meet compendial requirements”, or a footnote indicating the last measurement. This is the origin of the “Conforms” and “Complies” entries seen throughout the research chemical market. Those entries are not necessarily evasive. They may be exactly what the convention prescribes for an attribute legitimately not measured on this lot. Note also that the IPEC guide is voluntary, states that its contents are not regulatory requirements, and uses “should” rather than “must”, so an issuer who omits the annotation has broken no rule.

The model certificate in Annex 1 of the guide is the most instructive artefact in the document. It lists nickel with an asterisk footnote reading that the test is performed in-process on each batch, and arsenic with a separate footnote reading that the test is performed quarterly based on process validation. The reported result for arsenic is printed as “NMT 2 ppm”, identical to the specification. A reader scanning the results column sees a value that satisfies the limit. Only the footnote reveals that the value is the limit itself and that the underlying measurement may be up to three months old and made on different material. This is the reporting convention working as designed, and it is also precisely the failure mode when the footnote is dropped in transcription.

Residual Solvents: The Canonical Skip Candidate

ICH Q6A names residual solvents as a primary example of an attribute suited to periodic testing. The logic is sound in a validated process: if the manufacturing route is fixed and the drying step has demonstrated capability across many batches, the solvent profile is a function of the process rather than of the individual lot.

The logic weakens as process control weakens. Solid-phase peptide synthesis uses N,N-dimethylformamide as the principal coupling solvent, dichloromethane in washing and cleavage steps, acetonitrile throughout purification, and trifluoroacetic acid in cleavage and as a mobile phase modifier. ICH Q3C(R8) assigns acetonitrile a permitted daily exposure of 4.1 mg and a concentration limit of 410 ppm, dichloromethane 6.0 mg and 600 ppm, and N,N-dimethylformamide 8.8 mg and 880 ppm. These limits are calculated for a 50 kg reference adult and are not directly applicable to laboratory work, but they set the analytical thresholds a gas chromatography headspace method is validated against. The relationship between those limits and what a purity number can and cannot capture is covered in more detail in our discussion of residual solvents and gas chromatography headspace testing.

The critical dependency is process stability. IPEC Annex 2 sets out when skipping is defensible, and the central condition is that the process capability index on the relevant parameter is high and based on a stable process, meaning output that demonstrably meets all aspects of the stated specification. A contract synthesis operation running varied sequences on shared equipment with changing scales and raw material sources does not meet that description. Skipping a solvent test in that setting is not periodic testing under a justified plan. It is an untested attribute with a number beside it.

What Happens When a Periodic Test Fails

ICH Q6A is specific about the consequences. When a periodic test is performed and fails to meet the acceptance criteria established for it, the failure should be handled by proper notification of the appropriate regulatory authorities. If the data demonstrate a need to restore routine testing, then batch-by-batch release testing should be reinstated.

The implication is uncomfortable and rarely stated plainly. A failure on a periodic test is not evidence about one batch. It is evidence that the process assumption underwriting every unmeasured batch since the last successful test may be wrong. If a solvent test is run once in 20 batches and fails, the 19 batches released on the strength of the previous result were released on an assumption the new data has just contradicted. The investigation obligations that follow are the same ones that govern any anomalous laboratory result, which we cover in our analysis of out-of-specification results and retesting limits. In a regulated setting this triggers notification, scope assessment and potentially recall. In an unregulated one, nothing happens, because nobody outside the manufacturer learns the test was run.

Swissmedic’s 2018 workshop material illustrates the frequencies at stake, citing industry proposals of one batch tested in every 20 to 30 batches, or one per year, whichever came first. It also noted the PIC/S GMP Part II position that each batch incorporated into a blend should have been tested and that sub-batches with out-of-specification results must not be blended. Blending is directly relevant to peptide supply, where material from multiple synthesis campaigns is sometimes combined before vialling and the resulting certificate reports a single set of numbers for a composite whose components may not have been individually characterised. How two nominally identical lots diverge is covered in our piece on batch-to-batch comparability.

The Statistical Machinery Behind Skipping

Skip-lot testing is not a pharmaceutical invention. It originates in acceptance sampling theory, where the aim was reducing inspection burden on suppliers with proven quality histories. Dodge, H. F. (1955), “Skip-Lot Sampling Plan”, Industrial Quality Control 11(5), pages 3 to 5, introduced the first system, now designated SkSP-1. Perry, R. L. (1970), in a Rutgers doctoral thesis, developed SkSP-2, which pairs the skipping rule with a reference sampling plan applied to the lots that are inspected.

An SkSP-2 scheme is defined by a clearance number i, the count of consecutive conforming lots required before skipping begins, and a sampling fraction f, the proportion of lots inspected while skipping. A single non-conforming lot returns the scheme to normal inspection until i consecutive conforming lots are again observed. The whole construction assumes a continuing stream of lots from one stable process whose quality history is known to the party inspecting. A purchaser of research peptides has no visibility into the clearance count, no knowledge of the sampling fraction, no record of whether the scheme is in normal or skipping state, and no mechanism to learn that a non-conforming lot has reset it. The statistical guarantee belongs to whoever holds the lot history, and downstream it does not transfer.

Why This Matters for Research Peptide Certificates

Every safeguard described above depends on a regulatory relationship that the research chemical sector does not have. Skip testing under Q6A requires a marketing authorisation applicant or holder to justify the reduction, a regulator to assess and approve it, a specification footnote recording the approved frequency, and an inspectorate able to compare practice against approval. Research peptides sit outside that structure entirely. There is no authorisation, no assessor, no approved frequency and no inspection.

What survives the transplant is the reporting convention without the control system that justified it. A certificate reporting “Conforms” against endotoxin, or printing the specification limit itself as the result for elemental impurities, follows a format designed for a context where an approved skip plan sat behind it. In the research market there is frequently nothing behind it, and the convention makes an untested attribute look identical to a tested one.

What a Researcher Can Actually Check

Three document-level checks separate a measured attribute from an asserted one, and none need laboratory access.

The first is the presence of numeric data rather than conformance language. IPEC section 7.2 recommends reporting actual data and observations rather than non-specific “passes” or “conforms” statements unless the test is qualitative. An attribute that is quantitative by nature, such as water content, residual solvent concentration or an elemental impurity, but is reported as “Conforms”, is an attribute whose measurement basis is undisclosed.

The second is whether the reported value is distinguishable from the specification. When the result column reproduces the limit exactly, as in the IPEC model certificate’s arsenic entry, the result is not a measurement. Real measurements land somewhere inside the acceptance range and vary between lots. A results column that mirrors the specification column across multiple attributes indicates a certificate built from limits rather than from data.

The third is the test date relative to the batch manufacturing date. A test date preceding the batch manufacturing date is a definitive marker that the result came from elsewhere, and the absence of any test date makes the question unanswerable, which is itself informative.

Beyond the document, the only real remedy is scope disclosure: a statement of which attributes were measured on the specific lot, by whom, and on what date, with the report retrievable from the issuing laboratory rather than from the seller. Maple Research Labs publishes its available batch reports on the certificates of analysis page so that the tested attributes and the identity of the testing laboratory are visible before purchase rather than inferred from a formatted table.

The Structural Point

Skip-lot testing is not a loophole and it is not misconduct. It is a rational response to redundant measurement in a process whose behaviour is known, and the guidelines permitting it also bind it: prior justification, regulatory approval, a recorded frequency, mandatory notification on failure, and reinstated batch-by-batch testing when the data demand it. Every one of those bindings lives upstream of the certificate.

What reaches a purchaser is a single page that flattens measured results, in-process surrogates, continuously monitored parameters and periodically sampled attributes into one uniform table. The pharmacopoeias were explicit that full monograph execution was never the standard. The certificate format was never designed to communicate which category each line belongs to, and in a market with no regulator enforcing the annotation, it usually does not. Reading a certificate well means asking not only whether the number is true, but whether it is about the material in front of you.

A document that omits results altogether raises the same question from the other direction. The distinction between a certificate of conformance and a certificate of analysis turns on whether the supplier is reporting measured values or simply asserting that a specification was met.

For research purposes only. Not for human consumption. Not for diagnostic or therapeutic use.