From Score to Strategy: Turning Proficiency Testing Results into Lab Improvement

Published 6 August 2026 · By Dr Sofia Lindqvist, Scheme Manager, LGC AXIO Proficiency Testing

In most laboratories, the proficiency testing report follows a predictable path. It arrives, someone checks that the z-scores are inside plus or minus two, the report is signed and filed for the next assessment, and work continues. The requirement has been met. The opportunity has been missed.

Proficiency testing is the only routine mechanism that tells a laboratory how its results compare with everyone else measuring the same thing. Internal quality control confirms that you are consistent; it cannot tell you whether you are consistently wrong. Used properly, a sequence of PT rounds is one of the richest diagnostic datasets a technical manager has access to.

Reading the score properly

A z-score expresses the difference between your result and the assigned value, scaled by the standard deviation for proficiency assessment. Conventionally, an absolute value below two is satisfactory, between two and three is questionable, and above three is unsatisfactory. Three points are routinely overlooked:

• A satisfactory score is not a precise score. A z-score of 1.8 sits inside the acceptance criterion while indicating a substantial deviation from the consensus.
• The denominator matters. Where the standard deviation for proficiency assessment is set generously, poor performance can be masked. Always check how it was derived.
• How the assigned value was established changes its meaning. A value from a reference measurement or gravimetric formulation is independent of participant performance; a consensus value is not.

The information is in the trend

A single round is a snapshot with real statistical noise. A sequence of rounds is evidence. Plot your z-scores by analyte, by matrix and by analyst over two or three years and distinct patterns emerge:

1. Persistent one-sided bias. Scores that cluster at plus 1.2, plus 1.5, plus 1.1 are never a coincidence. Look at calibration standard preparation, purity assumptions, recovery correction and the reporting basis before anything else.
2. Widening scatter. Increasing variability across rounds points to instrument or column degradation, environmental control, or inconsistency between operators.
3. Matrix-specific failure. Good performance in clean matrices and poor performance in complex ones indicates an extraction or matrix-effect problem rather than a calibration problem.
4. Concentration-dependent error. Deviation only at the low end usually means a limit of quantification set optimistically, or blank contamination.
5. Analyte-family clustering. A shared error across chemically related analytes often traces back to one mixed standard or one shared internal standard.

None of these patterns is visible from a single report, and each one points to a different corrective action.

Investigation that actually finds the cause

When a score falls outside acceptance, the temptation is to repeat the analysis, obtain a better number and close the record. That confirms nothing except that the result is reproducible on a second attempt. A disciplined investigation works through the chain in order:

• Transcription and unit errors — still the single most common root cause, and the easiest to rule out.
• Reporting basis — dry weight versus as-received, acid versus anion, salt versus free base.
• Calibration standard integrity — lot, purity basis, expiry, preparation records, dilution arithmetic.
• Sample preparation — recovery, homogenisation, digestion completeness, spike placement.
• Instrumental performance on the day — sensitivity, carryover, interference, integration decisions.
• Only then: the possibility that the assigned value itself is questionable, which does occasionally happen and should be raised with the provider.

Building PT into the quality system

Laboratories that extract the most value treat proficiency testing as a planning input rather than a compliance output. In practice that means selecting schemes that genuinely cover the scope and concentration ranges you report, not just the convenient ones. It means scheduling rounds so they test the method under normal conditions rather than being handled by the most experienced analyst on a quiet afternoon. It means reviewing PT trends at management review alongside complaints and internal audit findings, and using them to justify investment in equipment, training or method redevelopment.

A quarterly routine worth adopting

1. Update a running z-score chart by analyte and matrix.
2. Flag any analyte with three consecutive same-sign scores, regardless of whether they passed.
3. Cross-reference PT deviations against internal QC charts for the same period.
4. Record one concrete technical action per round, even where all scores were satisfactory.
5. Feed recurring themes into the annual training and method development plan.

Why it is worth the discipline

Every proficiency testing round is an inexpensive, externally validated check on whether the results leaving your laboratory would be recognised as correct by your peers. For laboratories whose results determine whether food reaches a shelf, a medicine reaches a patient or water reaches a tap, that external perspective is not an administrative formality. It is the closest thing the sector has to an independent second opinion.