How to choose and instruct a laboratory for chemical food analysis

It will clearly help to think about what you want the laboratory to do, before approaching any laboratories for quotes and advice. The table below summarises common tasks that require laboratory services and some guidance. As a general principle: have a conversation with the lab, ensure they understand what they are doing and what you are doing.

Table 1: What do you want the laboratory to do and what you should think about?

The performance of official control laboratories is controlled by Regulation 2017/625 on official controls. While possibly more exacting that for a low-cost high throughput commercial laboratory the requirements set out in this regulation are a useful place to start.

These requirements are that the laboratory should have:

1. The expertise, equipment and infrastructure required to carry out analyses or tests

Accreditation, which requires participation in appropriate proficiency test rounds is a good indication of expertise. Check that the required equipment and space to carry out the claimed workload is available. Some applications (e.g. quick turn-round suites of pesticides analyses for ≥300 analytes, GMOs by PCR or allergens by LC-MS/MS) may require the detection capability provided by relatively recent versions of high-end instrumentation. Check that the scope of the method particularly multi-residue methods (e.g. pesticides and veterinary residues) includes the specific substances you are interested in. Check if the instrumentation continues to be supported by its supplier or manufacturer. On the other hand, robust well-maintained triple quadruple Mass Spectrometry (MS) instruments provide useful performance over a span of at least 10 years. A laboratory information management system is generally required. Consider if the scope of work the laboratory undertakes is compatible with your needs. As an example, if the laboratory specialises in trace metals analysis at low (μg/kg[1]) levels, and you send them a pre-mix sample containing 10% metals, then your sample will cross-contaminate their equipment.

2. Sufficient number of suitably qualified, trained and experienced staff

You can assume from ISO17025 accreditation that staff are suitably trained, however, you should assure yourself that the laboratory has sufficient competent staff to deliver your requirements, regardless of their competing priorities, and has adequate contingency and succession planning. Chartered Scientist (CSci) and Registered Scientist (RSci) status is awarded under license from the Science Council by IFST, for example, see this matrix.

Note that, unless separately listed, ISO17025 accreditation specifically excludes opinions and interpretations as it applies only to the factual test results.  Laboratories are not permitted to put interpretations on the CoA.  If you want your laboratory to explain the context of results to you, then it is important to look for scientists, or Key Account Managers, who are not only technically competent but also good communicators.

3. Performs impartially and free from any conflict of interest

Look for evidence of commitment at a senior level, such as Board and Managing Director.

4. Can deliver in a timely manner

Decide on your requirements in terms of product recall (safety), marketing timescale, new product launch, when you really need it, add in time for transit and to discuss results, processing data and making sense of it. Integrate your critical deadlines with those of the laboratory and discuss with them. Remember that analytical runs must include quality control (QC) samples (including replicates, controls, blanks) and a minimum number of billable samples (typically 10, as a minimum) must be in the run for it to be financially viable. Thus, an urgent ‘one-off’ will usually be more expensive. Laboratories often have a routine cost and a fast track cost. Owing to the required analytical protocol, e.g. stipulation of overnight hydrolysis to release the analyte of interest, methods may have a minimum time of completion. Quality control assessment of a run is essential and adds time. Laboratories often have, and should have, delegated levels of authority to release results, which may add to the time taken to report.

5. Operates in accordance with ISO/IEC 17025 standard and is accredited in accordance with that standard, by a national accreditation body

As noted above, ISO/IEC 17025 accreditation[2], or GLP[3] is a key attribute. The scope of the accreditation should include both your sample type and the proposed test method(s), although you may need to ask the laboratory for advice on this. The laboratory may have a flexible scope of accreditation which covers a group of related methods and, with certain conditions, may be applied to an analyte/matrix combination for which the laboratory does not hold specific accreditation.

You may need a test for which the lab has a method/protocol for but do not have accreditation. The Quality System in place must still cover such work and you might seek an assurance that this is so. Moreover, the lab should be able to demonstrate the method is validated and has acceptable performance characteristics.  You may wish to check the reason why a lab chooses to withdraw a method from the scope of accreditation as they may have been in danger of losing it owing to deficits in performance. Note however that since accreditation is a significant overhead withdrawal may have been solely for commercial reasons. Even if unaccredited the results must be properly reported.

A glance at the table of contents of ISO/IEC 17025:2017(en) ‘General requirements for the competence of testing and calibration laboratories’[4] shows that most of the aspects addressed in this IS are covered by this laboratory standard. These include impartiality and confidentiality, resource requirements (personnel, equipment and ‘subcontracting’), process requirements including contract review, selection of methods, sampling, sample handling, records, the method performance characteristics (see below), reporting results, handling complaints, the quality management system itself, and metrological traceability.

A key requirement of ISO/IEC 17025 is that laboratories communicate with you in advance, to agree on how they will account for measurement uncertainty when giving a compliance decision against any limit or specification.  This topic is covered in IFST Information Statement: ‘How to Interpret Laboratory Results’

The international organisation for accreditation bodies is ILAC. The UK accreditation body is UKAS[5].  You can search for a lab that carried the required UKAS accreditation here. Other national accreditation bodies have similar searchable databases.

Campden Laboratory Accreditation Scheme (CLAS) is an independent laboratory accreditation scheme, for food industry laboratories, established specifically to meet the need for a recognised standard for food, drink and allied laboratories[6].  It is a less detailed standard than ISO17025, with consequentially lower overheads. It is particularly suitable for in-house laboratories, where you do not need international mutual acceptance of results, but still want an external stamp of approval.

Chemical and bio-analytical results cannot be perfect; this should not come as a big surprise. We use the term measurement uncertainty to describe this lack of perfection[1].

Thus, in the reporting of results (in certificates of analysis and in scientific papers) an expression of the following form can be found:

Result of analysis:

Parameter name (e.g. Aflatoxin B₁) … x ± y followed by a unit

This is the mean ± the uncertainty.

Measurement uncertainty includes both random and systematic effects (see Appendix 1). If there are significant known biases, then the laboratory can adapt the method to try and remove or reduce them. However, there will always be systematic effects present including, for example, in relation to the calibration of equipment, and tolerances associated with glassware.

The standard uncertainty (u) represents about 68 % of the possible spread of results in a ‘normal’ (Gaussian) distribution, as in the familiar ‘bell-shaped curve’. Laboratories usually report the expanded uncertainty where ‘u’ is multiplied by a ‘coverage factor’[2] ‘k’.  This is typically set as equal to 2, which gives an expanded uncertainty representing over 95% of the possible spread of results. Best practice for laboratories is to include expanded uncertainties on their certificates of analysis, and these should be accompanied by a statement of the level of confidence and the coverage factor used. However, in practice, laboratories may report in a number of ways. The uncertainty can be a standard deviation ‘s’, the standard uncertainty ‘u’, or the expanded measurement uncertainty ‘U’.   It is important that you understand which they use. The practical implications of this are:

• if you want to know the uncertainty, you may need to ask the laboratory to state the measurement uncertainty as it may not be routinely reported
• if you want to understand the uncertainty, you will need to check with the laboratory which of the above measures has been reported: standard deviation (s), standard uncertainty (u), or expanded measurement uncertainty (U)
• if you want to compare results from different laboratories, or from different methods of analysis, you can only do so with a knowledge of the measurement uncertainty
• the lower, or occasionally the upper, bound of the measurement uncertainty may be the datum of interest to establish whether the result is compliant or otherwise.

When a sample is re-tested, it often prompts the question ‘why are the two results different?’ or even, ‘are the two results different?’.

There are statistical procedures to assess whether two results are comparable. Chapters on tests of significance and ‘the null hypothesis' can be consulted in standard works (see above). For the purposes of this Information Statement, however, a simple approach by inspection is probably sufficient. Illustrated graphically in Figure 2 for four results for equivalent samples.

Figure 2: Four results for equivalent samples; the y axis is arbitrary concentration units

If it can be assumed that the samples were truly equivalent, as per the responsibility of the sampler, and results ‘A’ and ‘B’ can be considered comparable since their error bars overlap. Result ‘C’ differs sufficiently from either ‘A’, ‘B’ or ‘D’, to warrant further investigation. Result ‘D’ overlaps ‘A’ and ‘B’, however, its measurement uncertainty is larger, and it may be suboptimal compared to them for that reason.

Other considerations

Outside of routine analyses, well understood by you the customer, it may be necessary to explore the degree to which your chemical and bio-analytical service laboratory understands the context of what they are doing for you so that they can properly advise you, and offer a valid interpretation of results. Examples are given below.

Flawed sampling wastes time and money, both in the sampling and subsequent analysis. Laboratories should be able to, and usually can give advice on sampling, sample handling (e.g. containers, storage temperatures), and the correct order of subsampling, if multiple tests are required. See IFST Information Statement ‘Sampling for Food Analysis - key considerations’. The location of your laboratory, how best to package and transport the samples (so they arrive in a condition to validly test), and whether or not they have a logistics arrangement to pick up samples may be important. A record of the chain of custody of the sample may be necessary if subsequent legal proceedings or litigation is in prospect.  

In the EU and UK, official food laboratories are ‘Public Analysts’ and ‘Food Examiners’ (microbiology laboratories) that undertake work for local authorities in support of enforcement of food standards and food safety. Other designated laboratories undertake official controls in specified areas. All official laboratories must employ staff who have qualifications defined by national legislation.  In addition to be named an official control laboratory, laboratories must have a ‘Public Analyst’, formally appointed by a local authority, or a qualified ‘Food Examiner’.

In the UK, the Food Standards Agency (FSA) and Food Standards Scotland (FSS) are responsible for appointing[1] the majority of official food control laboratories. FSA and FSS do not own or operate any of these laboratories. Some are private sector organisations, and some are owned and operated by Local Authorities. Many carry out commercial work in addition to their official duties. There are statutory requirements for qualifications and experience, infrastructure, accreditation and avoidance of conflict of interest[2],[3].

Non-targeted analysis (NTA) reports previously unknown or poorly studied compounds, for example by liquid chromatography-tandem mass spectrometry datasets of isotopic masses, retention times, or predicted molecular formulae. NTA is particularly useful to explore non intentionally added substances and although conventional validation studies are difficult, should be attempted[1].

Vitamins analysis is often a specialist activity and an experienced accredited laboratory is to be preferred.