Common NABL non-conformities and how to avoid them

How do NABL assessors find non-conformities?

Assessors find non-conformities by tracing real work and witnessing tests. NABL 100B says the assessment team checks that the documented system is implemented and effective, and that the lab's technical competence is "evaluated and witnessed during the assessment".

In practice an assessor picks an issued report and walks it backwards: report, review, worksheet, raw data, reagent standardisation, equipment, analyst authorisation, sample receipt, contract. Each link must exist and agree with the others. A gap anywhere in that chain becomes a finding.

NABL 100B then gives the lab 30 days to submit corrections and corrective actions based on root cause analysis, with documentary evidence.

Why are technical records the most exposed area?

Technical records carry every result, so any weakness in them affects everything else. ISO/IEC 17025 clause 7.5.1 requires original observations, data and calculations to be recorded at the time they are made, with the date and the person responsible for each activity and for checking. Clause 7.5.2 requires amendments to be traceable to the original, with both versions kept.

Assessors often look at:

• titres or readings copied later from rough paper into a fair register;
• overwriting or correction fluid instead of a single line with initials and date;
• Excel sheets where a result cell was retyped and the earlier value is gone;
• no record of who checked the calculation.

What traceability problems do assessors look for?

Assessors look for an unbroken chain from each result to SI units or a stated reference. Clause 6.5 and NABL 142 require calibration by a competent laboratory, or certified reference materials from a competent producer.

Typical gaps they check:

• calibration certificates from a lab whose scope does not cover the instrument or range used;
• reference standards or CRMs without a valid certificate, or used after expiry;
• volumetric glassware, thermometers and weights with no calibration record;
• working standards prepared in the lab with no link back to a certified material.

What do assessors check on method validation and verification?

Assessors check that every method in the scope was verified before use, and that non-standard or modified methods were validated. Clause 7.2.1.5 requires a lab to verify that it can perform a standard method and to keep records; clause 7.2.2 requires validation for non-standard, lab-developed or modified methods.

Common weak points include:

• verification done once on an older edition, then not repeated when the standard was revised;
• a "standard" method run with a changed sample volume or reagent, with no validation;
• LOD and LOQ quoted in the scope but not supported by data;
• clause 7.2.1.3 not met because the lab still uses a superseded edition.

What goes wrong with measurement uncertainty?

What goes wrong most is an uncertainty estimate that exists on paper but does not reflect how the method is run. Clause 7.6.3 requires a testing lab to evaluate measurement uncertainty, and NABL 141 is NABL's guideline for doing it.

Assessors often look at:

• budgets missing major contributions, such as sampling or recovery;
• the same uncertainty used across very different concentration ranges;
• uncertainty not considered when a result sits close to a limit and the report says "complies" (clause 7.8.3.1).

How is calibration status checked?

Calibration status is checked at the bench, not only in the register. Clause 6.4.8 requires equipment to be labelled or coded so users can see its calibration status; clause 6.4.9 requires faulty or overdue equipment to be taken out of service; clause 6.4.13 lists the records to keep.

The classic finding: the register shows a balance overdue, but the worksheet from last week records weighings on it. A paper register cannot stop that. Intermediate checks (clause 6.4.10) without a written procedure, and records that miss the next due date, are also common. Our guide on instrument calibration in a testing lab covers the calibration programme itself.

Why do sample receipt and holding times cause findings?

They cause findings because evidence is lost at the busiest desk in the lab. Clause 7.4 requires a procedure for receipt, handling, storage and disposal, unambiguous identification, and recording of any deviation found at receipt. When the item is unsuitable, the lab must consult the customer and, if testing goes ahead, state in the report which results may be affected.

Assessors often look at:

• the date and time of sampling missing, so nobody can show the method's holding time was met;
• a sample received short or warm, tested anyway, with no customer consent and no note on the report;
• sample codes written by hand that do not match the worksheet;
• storage temperature records with gaps on Sundays and holidays.

Which report errors turn into findings?

Report errors that turn into findings are usually missing clause 7.8.2.1 content or misuse of the NABL symbol. Examples assessors often look at include:

• no statement that results relate only to the items tested;
• no date of analysis, or a date that does not match the worksheet;
• "complies" printed with no decision rule documented (clause 7.8.6);
• non-accredited parameters printed under the NABL symbol, or marked with an asterisk, which NABL 133 does not permit;
• an amended report that does not identify the change and the reason (clause 7.8.8).

Our guide to the NABL test report format covers each rule.

What are impartiality findings about?

Impartiality findings are about risks the lab has not identified or controlled. Clause 4.1.4 requires the lab to identify risks to impartiality on an ongoing basis, from its activities, relationships and staff; clause 4.1.5 requires it to show how they are removed or reduced.

Assessors often ask how the lab stops commercial pressure reaching the analyst, for example when a large customer's results are near a limit. A risk register that was filled once and never reviewed is a common weak point.

What is "control of data" and why does Excel fail it?

Control of data is clause 7.11: any system that collects, processes, records, reports or stores data must be validated before use, protected from unauthorised access, safeguarded against tampering and loss, and have its calculations and data transfers checked. Note 1 to clause 7.11.2 says this covers non-computerised systems too.

A shared Excel workbook fails it in predictable ways: formula cells anyone can edit, no record of who changed what, one copy on one PC, and no tested backup. Clause 8.4.2 adds control of back-up, archive and retention for records. Our guide to the hidden cost of Excel in labs looks at this in detail.

Registers and Excel compared with a controlled system

The table below maps each area to the clause, the gap that manual methods tend to leave, and what closes it, whether paper-based or software-based.

A NABL audit checklist for your own mock assessment

Before a NABL assessment, run your own trace on a few recent reports. NABL publishes its forms for this: NABL 219 (assessment forms and checklist) and NABL 220 (document review checklist), both based on ISO/IEC 17025:2017.

1. Pick three issued reports from different sections, including one amended report.
2. For each, find the contract review and the sample receipt record, with date and time of sampling.
3. Check the holding time was met, or that a late start is reported.
4. Open the worksheet and raw data; recalculate one result by hand.
5. Check the analyst was authorised for that test on that date.
6. Check every instrument used was within calibration on that date, with a traceable certificate.
7. Check the reagent standardisation and QC result for that batch.
8. Check the uncertainty estimate and decision rule behind any "complies".
9. Check the report against clause 7.8.2.1 and NABL 133.
10. Check you can restore the data from last month's backup.

How should a lab respond to a NABL non-conformity?

A lab should respond with correction, root cause and corrective action, backed by evidence, within the 30 days NABL 100B allows. Clause 8.7 of ISO/IEC 17025 sets the steps: control and correct the problem, find its cause, check whether similar problems exist elsewhere, act, and review whether the action worked.

"Retrained the analyst" alone rarely closes a finding, because it does not show why the error happened or that it cannot recur. Show the changed procedure, form or system, and records made since the change.