"Third-party tested" and "accredited" are frequently used as though they were a single guarantee. They are neither single nor a guarantee. Accreditation is a formal judgement about a defined list of activities at a defined laboratory; validation is a separate exercise carried out on an individual analytical procedure; and an independent test result describes one sample on one date. This page separates the three and sets out what each contributes.
What accreditation to ISO/IEC 17025 covers
ISO/IEC 17025 is titled as general requirements for the competence of testing and calibration laboratories, and it is the standard against which testing laboratories are assessed in formal recognition programmes 1. The standard sets requirements for laboratories to demonstrate competency and the ability to generate valid results 2.
Its content falls into two groups. In the version examined in a published case study of an accredited national medicines-control laboratory, the standard's twenty-five clauses divide into management requirements, covering the legal, financial and other administrative obligations of running a laboratory, and technical requirements, covering the activities that determine how tests are conducted 2. Compliance is assessed by audit, and shortfalls are recorded as non-conformities against individual clauses 2.
Accreditation is granted by an accreditation body, not by the laboratory itself and not by a customer. Where a regulator recognises those bodies it applies its own criteria to them: under the FDA's food-testing accreditation programme, a recognised accreditation body must be a full member of the International Laboratory Accreditation Cooperation, a signatory to its mutual recognition arrangement, and itself competent to ISO/IEC 17011 with a scope of "Testing: ISO/IEC 17025" 3. Accreditation is therefore a layered arrangement, and each layer has its own scope.
The scope of accreditation is the operative document
A laboratory's accreditation certificate says that it was assessed. The scope of accreditation says what it was assessed for.
The scope is finite and enumerated, and it changes over time: the laboratory described in the published case study held accreditation in eleven test scope areas in 2015 and sixteen areas by 2018 2. A scope also carries exclusions. In the FDA's medical-device programme the relevant section of the scope issued by the accreditation body is expected to list any exclusions explicitly, such as exclusions from particular clauses of a given standard 4.
An accredited laboratory is not an accredited method
This is the distinction most often lost. Accreditation attaches to named methods and standards within a scope; it does not travel with the laboratory to everything else it is capable of running.
The rule is stated plainly in the FDA programme: a testing laboratory may label its testing as having been conducted under the programme only if the standards and test methods used were within its scope of accreditation at the time of testing 4. Two conditions are embedded there. The method has to be inside the scope, and it has to have been inside the scope on the date the test was run. The same document notes that accreditation under the programme is separate from any other accreditation status the laboratory holds with the same body 4.
A certificate can therefore come from a genuinely accredited laboratory and still report a result produced by a method outside that laboratory's scope. Nothing about that is improper, but it is a different claim, and it is only visible if the method and the scope are both available to be compared.
What method validation involves
Validation is the demonstration that a particular analytical procedure is fit for its intended purpose, and the tests that make up that demonstration are enumerated in ICH Q2(R2).
Specificity and selectivity describe the extent to which other substances interfere with the determination of an analyte 5. They can be shown by demonstrating the absence of interference, or by comparing results against an orthogonal procedure; for a critical chromatographic separation, specificity can be demonstrated by the resolution of the two components eluting closest to each other 5. Where one procedure cannot discriminate sufficiently, the guideline recommends combining two or more 5.
Range and response cover the interval between the lowest and highest results for which the procedure has a suitable level of response, accuracy and precision 5. Where the relationship between concentration and signal is linear, a minimum of five concentrations appropriately distributed across the range is recommended, evaluated by an appropriate statistical method such as least-squares regression, with the plot, correlation coefficient, intercept and slope reported 5.
Accuracy expresses the closeness of agreement between the measured value and an accepted reference value 5. It is established across the reportable range under the regular test conditions of the procedure, typically by applying the procedure to an analyte of known purity such as a reference material, or by a spiking study 5.
Precision expresses the degree of scatter among repeated measurements of the same homogeneous sample, and is considered at three levels: repeatability under the same conditions over a short interval, intermediate precision covering intra-laboratory variation across days, analysts, equipment and conditions, and reproducibility between laboratories 5. Standard deviation, relative standard deviation and a confidence interval are the reported quantities 5.
Detection and quantitation limits are estimated where the attribute being measured brings the range close to the lower limits of the procedure 5. The detection limit is the lowest amount that can be detected but not necessarily quantitated as an exact value; the quantitation limit is the lowest amount that can be quantitatively determined with suitable precision and accuracy, and is the parameter that matters for impurities and degradation products 5. Both may be estimated visually, from a signal-to-noise comparison against blanks — with a ratio of 3:1 generally acceptable for the detection limit and at least 10:1 for the quantitation limit — or from the standard deviation of the response and the slope of the calibration line 5.
Robustness is a measure of the capacity of a procedure to meet its expected criteria during normal use, and it is tested by deliberate variations of the procedure's parameters 5. It is usually evaluated during development, before the validation study is executed 5.
Transfer of a validated procedure to a different laboratory is itself treated as a change: partial or full revalidation, or comparative analysis of representative samples, is expected 5.
What independent testing does and does not establish
What it does establish is narrow and real. An independent result is generated by a party with no interest in the outcome, using a stated method, on a stated sample, on a stated date.
What it does not establish follows from the same sentence. It says nothing about material other than the sample tested. It says nothing about methods outside the scope under which the laboratory was assessed 4. And it does not by itself make one laboratory's figure comparable with another's: reference-standard work that assigns values across multiple laboratories weights them by their variance and identifies outliers statistically, precisely because within-laboratory variability differs between laboratories 6. Accreditation constrains how a result was produced; it does not convert a result into a property of a product.
How to read a scope
Four questions do most of the work when a scope document is in front of a reader.
Which body issued it, and is that body itself recognised under a mutual recognition arrangement 3. Which standard was assessed, and in which version 2. Which test areas and methods are listed, and which exclusions are stated 4. And whether the method that produced the result in hand appears in that list, and appeared in it on the date of the test 4.
A scope is a list, not a badge. The useful comparison is always between the method named on the certificate and the methods named in the scope, read for the date the test was run.