Third-party testing: who does what

What a third party is for

The argument for independent testing is simple: a supplier reporting on its own material has an interest in the answer, and an outside laboratory does not. That is a genuine improvement and it is the reason this whole ecosystem exists. It is also a narrower improvement than the phrase third-party testing implies, because independence of the laboratory is only one of several places bias can enter.

Consider the sequence. Somebody chooses which product to test. Somebody chooses which vial. Somebody packs and sends it. Somebody chooses which analyses to pay for. Somebody decides whether to publish the result. A laboratory sitting in the middle of that chain can be perfectly impartial about the measurement while every decision around it was made by an interested party.

Sunil holds the quality-and-testing circle and puts it as a question rather than a rule: independent of what, exactly? Answering that for a given report is most of the analysis.

The names members will encounter

Four names come up constantly in this community, and members are often unclear that they do different jobs.

Janoshik is an analytical laboratory. It receives samples and produces reports — typically purity by chromatography and identity by mass spectrometry, sometimes quantity. Reports carry sample identifiers and dates. Both suppliers and individuals commission work from it, which matters enormously, because the same laboratory name appears on a supplier-commissioned report and on one a member paid for out of their own pocket, and those two documents mean quite different things.

Medutest and PeptideMeter operate as testing and publishing programmes: material is obtained, analysis is arranged, and results are published against a named product. The value here is partly the publication — a programme that only ever printed favourable results would be a marketing channel, and the reason members here take PeptideMeter seriously is that it has published results a supplier would plainly rather it had not.

VendorInvestigate is an aggregation and investigation surface rather than a laboratory. It collects listings, reports and findings across suppliers. Aggregation is genuinely useful for spotting a pattern over time and is not a substitute for a measurement.

None of these names is endorsed here, and we have no relationship with any of them.

Who paid changes what the report means

Take a single purity figure and vary only the commissioner, and watch the meaning move.

The supplier commissioned it. The supplier chose the product, chose the vial, sent it, and chose whether to publish the result. The measurement may be impeccable. The selection is not observable. Assume that the batches you see reports for are the ones that produced reports worth showing.

A publishing programme commissioned it. Better, because selection sits with someone whose reputation depends on publishing whatever comes back, including bad results. Still one vial, still bought at one time, and a supplier can behave differently towards a buyer it recognises.

A member commissioned it on their own vial. The strongest form available to individuals, because the sample chain runs from the vial that arrived at their house to the laboratory. It also has the narrowest scope: it tells you about that vial and nothing about the next one.

Members have done this — two people in our circles sent material from the same supplier and both came back within a couple of percentage points of the published figures. That is reassuring, and it is two vials out of an unknown number of batches, which is the sentence that has to follow.

Accreditation, and what it promises

The international standard for testing laboratories is ISO/IEC 17025, and accreditation to it is a real thing that a real assessment body grants after examining methods, equipment, calibration, staff competence and quality systems. It is meaningful.

Two qualifications keep members out of trouble. First, accreditation is scoped: a laboratory is accredited for specified methods on specified matrices, and work outside that scope is not covered by the accreditation even though the same building and the same staff perform it. A report should state whether the reported test was within scope; many do not. Second, accreditation speaks to competence and consistency, not to the significance of the result. A perfectly executed purity determination on an unrepresentative sample is a perfectly executed measurement of an unrepresentative sample.

It is also worth knowing that laboratories serving this market are not generally operating under medicines regulation, because they are analysing research chemicals rather than medicinal products. That is not a criticism of the laboratories. It is a description of where they sit.

Reference materials and the comparison problem

Identity is established by comparison. A laboratory says a substance is what it claims to be by measuring it against something already known to be that substance — a certified reference material with documented identity, assigned purity and a stated uncertainty. Sigma-Aldrich is one of the established suppliers of such materials, and access to appropriate standards is part of what separates a defensible identification from a plausible one.

The reader’s question is therefore: against what was this compared? For mass spectrometry, a theoretical mass is a calculation and a good check, but a retention-time match means nothing without a standard run on the same system. For anything reported as quantity, a calibration against a reference standard of known content is the entire basis of the number.

Certified reference materials for newer analogues are not always available, and where they are not, laboratories work with what exists. Reports rarely spell this out, and it is a fair thing to ask about.

Commissioning your own, honestly

Members do it, and it is the most informative thing an individual can do. It is worth being clear about what it buys.

You control the chain of custody from your fridge to the laboratory, which removes the biggest source of selection bias. You choose the analyses. You get an answer about the material you are actually going to use. Against that: it costs money that could have gone on the medicine itself, it consumes a vial or part of one, transport conditions are now your responsibility, and turnaround is measured in weeks, by which time you have usually made your decision anyway.

And the scope stays narrow. One vial tested is one vial known. A purity and identity panel still says nothing about sterility, endotoxin, residual solvents, elemental impurities, fill accuracy or storage history — the same list as always, and it does not shorten because you paid for the test yourself.

Badges, panels and outsourced judgement

The failure mode we see most is not credulity about a specific report. It is relief. Somebody finds a supplier with an apparently comprehensive testing presence, feels the anxiety drop, and stops asking questions — and the relief, not the report, is what the testing surface was designed to produce.

A verified badge on a shop’s own page is a graphic. A logo is not an accreditation. A test panel covering identity and purity is two questions answered out of a dozen that matter. And a programme’s coverage of a supplier tells you that somebody tested something once, at a time, from a channel you cannot see.

The useful posture is boring and it works: read the report for what it measured, note who chose the sample, and keep the absent list in view. Everything on that list stays open no matter how many laboratories are involved, and pretending otherwise is how a document ends up doing a job that only a regulator could do.