Reading a Certificate of Analysis
What each section of a certificate of analysis is asserting, and how to tell a document that reports measurements from one that reports adjectives.
6 min read
What the document is
A certificate of analysis reports the results of testing performed on a sample of a specific lot. It is a record of measurements, issued by whoever performed them, describing the material identified on it.
That scope is narrower than it is often taken to be. The document reports what was measured on the sample tested. It does not certify a production process, and it says nothing about material outside the lot it names.
Identification
The header establishes what the document refers to. It should name the compound, the lot number, and where applicable the batch number, and it should carry a date of analysis and a date of issue.
The lot number is the load-bearing field. Without it the document cannot be connected to any particular container, and every result it contains becomes unverifiable. A certificate that names a compound but not a lot is not traceable to anything.
Methods and results
Each result should be accompanied by the method that produced it. A purity figure should state that it was determined by HPLC. An identity statement should state that it was determined by mass spectrometry. Where conditions materially affect interpretation, those should be available as well.
Results without methods are the most common weakness in a weak certificate. A percentage with no stated method cannot be interpreted, because the reader has no way to know what was measured or what the figure is a proportion of. Two documents reporting the same number from different methods are reporting different things.
Supporting data strengthens a certificate considerably. A chromatogram lets a reader see peak shape and resolution rather than taking an integrated number on trust. A mass spectrum lets a reader compare the observed mass against the expected value directly. Summary figures alone ask for more confidence than they earn.
Attribution
The document should identify who performed the analysis. Testing may be conducted by the manufacturer or by an independent laboratory, and the distinction is a fact about the record that a reader is entitled to know.
Where a laboratory is named, the name should be specific enough to be checked. An unattributed certificate reports results without anyone standing behind them.
Signals worth weighing
Measurements can be checked; adjectives cannot. A document reporting a numeric result with its method is making a falsifiable statement. A document describing material as high quality or premium grade is not reporting an analysis, because those terms correspond to no measurement.
Other things worth noticing: whether the lot number on the document matches the container, whether dates are present and internally consistent, whether supporting data is available rather than only summary values, and whether fields that have no verified value are marked as pending rather than filled with something plausible.
The last of these is the most telling. A document that leaves a gap visible is describing the state of the evidence. A document with no gaps at all may simply be one where the gaps were filled in.
This article is general reference material describing methods and records. It does not describe any particular material, contains no dosing, reconstitution, or administration information, and is not medical advice.