Proficiency testing (PT) is the mechanism by which an accredited or certified laboratory demonstrates, using samples of unknown value sent by an outside provider, that it can produce results comparable to its peers. In clinical laboratories the same activity is often called external quality assessment (EQA) — the terms describe the same underlying practice, and ISO/IEC 17043 itself notes that “external quality assessment” is simply the name some medical-laboratory schemes use for their proficiency testing programs. Unlike internal quality control, which a laboratory runs on its own known materials, PT/EQA is run by a third party and graded against the performance of every other participating laboratory, which is what makes it the credible, comparable evidence that accreditation bodies and regulators rely on.
PT is not optional paperwork attached to accreditation — it is a recurring, scheduled obligation that has to keep being satisfied for the life of the accreditation or certification, not just at the point it is granted. This guide covers how PT schemes are structured under ISO/IEC 17043, how CLIA mandates it for U.S. clinical laboratories, how performance is scored (including the z-score convention used across most chemistry and quantitative schemes), what happens when a laboratory performs unsatisfactorily, and how PT differs in practice from routine internal quality control.
The two regulatory anchors: ISO/IEC 17043 and CLIA
ISO/IEC 17043 is the international standard, Conformity assessment — General requirements for the competence of proficiency testing providers, jointly published by ISO and IEC. It was revised in 2023 (ISO/IEC 17043:2023), replacing the original 2010 edition. The standard does not certify individual laboratories directly — it sets the requirements that a PT provider (the organization that designs the scheme, distributes samples, and statistically evaluates results) must meet to run a credible scheme. A laboratory’s own accreditation body (for example, in the context covered in our ISO/IEC 17025 guide, bodies such as A2LA, ANAB, NVLAP, or UKAS) requires the laboratories it accredits to participate in PT schemes that themselves meet ISO/IEC 17043, closing the loop between the two standards.
In the United States, participation in PT is a distinct, codified legal requirement for clinical laboratories under the Clinical Laboratory Improvement Amendments of 1988 (CLIA). The specific requirements sit in 42 CFR Part 493, Subpart H (“Participation in Proficiency Testing for Laboratories Performing Nonwaived Testing”), which lists the specialties, subspecialties, and analytes for which a laboratory holding a CLIA certificate of compliance or accreditation must enroll in a CMS-approved PT program, how many testing events per year are required, and how the laboratory must handle the PT samples (treating them exactly as it would treat a patient specimen, not with special handling that would not reflect routine practice). PT is, in fact, the only external, outcome-based performance indicator CLIA requires — personnel qualifications and internal quality control are also required, but PT is the one that compares a laboratory’s actual results against independently graded truth.
How a PT scheme actually works
The mechanics are consistent across most schemes, whether run by a professional body, a national metrology institute, or a commercial PT provider:
- Sample distribution. The provider sends every enrolled laboratory an aliquot of the same material — a homogeneous, stability-tested sample whose true or reference value the participating laboratories do not know in advance.
- Blind analysis under routine conditions. The laboratory analyzes the PT sample using its normal method, normal staff, and normal equipment, exactly as it would a real specimen — CLIA explicitly prohibits sending the sample to another laboratory for confirmation or analysis, referring to this as PT referral, which is treated as a serious violation in its own right.
- Reporting. Results are submitted to the provider by a fixed deadline.
- Statistical evaluation. The provider compares every laboratory’s result against an assigned or consensus value (often the overall group mean, a peer-group mean for laboratories using the same method, or a value assigned by a reference laboratory) and calculates a performance score for each participant.
- Reporting back. Each laboratory receives its own score, usually alongside anonymized summary statistics for the whole participant group, so it can see not just whether it passed but how its result compares to the distribution of peers.
Z-scores and how PT performance is evaluated
Most quantitative PT schemes (chemistry, environmental testing, metals analysis, and similar) score performance using a z-score, a standardized measure of how far a laboratory’s reported result falls from the assigned value, expressed in units of the scheme’s standard deviation for proficiency assessment:
z = (x − X) / σ
where x is the laboratory’s reported result, X is the assigned (reference) value, and σ is the standard deviation for proficiency assessment set by the scheme — a fixed target value chosen by the provider to reflect a fitness-for-purpose level of variation, not simply the raw spread of whatever results happen to come in that round. The statistical basis for this approach is set out in ISO 13528, the companion standard ISO/IEC 17043 references for the statistical methods PT providers use.
The conventional interpretation bands, widely used across PT schemes though the exact thresholds are set by each individual provider, are:
- |z| ≤ 2 — satisfactory performance.
- 2 < |z| < 3 — questionable/warning performance, not itself a failure but a signal worth investigating.
- |z| ≥ 3 — unsatisfactory/unacceptable performance.
Other scoring approaches exist for schemes where a strict numerical result isn’t the right fit — qualitative schemes (presence/absence, identification, typing) are typically scored simply as correct/incorrect or by percentage agreement with the reference identification, and some schemes use alternative statistics such as the En number (common in calibration and measurement comparison PT, comparing results with their associated measurement uncertainty) rather than a z-score.
What happens on unsuccessful performance
The consequences of an unsatisfactory PT result differ depending on which framework a laboratory is operating under, but the underlying logic is the same: an isolated miss triggers investigation, and a pattern of misses triggers escalating consequences.
Under CLIA (42 CFR Part 493, Subpart H), “unsuccessful performance” has a specific regulatory definition: failing to achieve the minimum satisfactory score for a given analyte, test, subspecialty, or specialty in two consecutive testing events, or two of three consecutive testing events. A single failed event on its own is not unsuccessful performance under CLIA’s definition, though most laboratories investigate every failed event regardless, as good practice and as the input to root-cause analysis. Once a laboratory does meet the regulatory definition of unsuccessful performance, it is required to:
- Cease reporting patient results for that analyte, test, subspecialty, or specialty until the problem is identified and corrected.
- Investigate the cause (instrument malfunction, method problem, transcription or clerical error, reagent lot issue, and so on) and document corrective action.
- Demonstrate two consecutive successful PT events (at least one of which may be conducted on site by the accrediting or oversight body) before the analyte can be reinstated.
Repeated or unresolved unsuccessful performance can lead to CMS sanctions under 42 CFR Part 493, Subpart R, up to and including suspension, limitation, or revocation of the laboratory’s CLIA certificate — which, in practice, means the laboratory cannot legally report results for the affected testing.
Under ISO/IEC 17025 accreditation, there is no single fixed regulatory penalty structure equivalent to CLIA’s — instead, an unsatisfactory PT result is treated as a nonconformity that the laboratory’s own quality management system must capture, investigate as a potential cause of invalid results, and correct through the same corrective-action process used for any other nonconformity, with the accreditation body reviewing the laboratory’s response (and, in serious or repeated cases, able to suspend or withdraw accreditation for the affected scope) at the next surveillance or reassessment visit.
How PT/EQA differs from internal quality control
These two activities are often confused because both are described as “quality,” but they test different things and neither substitutes for the other:
| Proficiency Testing / EQA | Internal Quality Control (QC) |
|---|---|
| Sample value is unknown to the laboratory at the time of testing | Control material’s expected value/range is known in advance |
| Run by an external, independent provider | Run by the laboratory itself, on its own schedule |
| Compares the laboratory against its peer group or an assigned reference value | Compares the laboratory’s result against its own historical performance |
| Typically a handful of events per year, per analyte/test | Typically run with every batch, every day, or per manufacturer instructions |
| Detects problems relative to the wider field — e.g., a method-specific bias other laboratories don’t share | Detects day-to-day drift, reagent or calibration problems, and short-term instrument instability |
| A regulatory/accreditation requirement in itself | Also a regulatory/accreditation requirement, but evaluated on different criteria |
The two are complementary rather than redundant: a laboratory can pass every internal QC check and still fail PT if its method has a systematic bias that its own control materials don’t reveal, and a laboratory can have an isolated internal QC failure on a given day without that ever showing up in its (much less frequent) PT results. Accreditation bodies and regulators require both because each one exposes a different failure mode.
What accredited laboratories should have in place
- A documented PT/EQA participation plan covering every accredited test/analyte, mapped against the applicable requirement (CLIA’s Subpart H list, the accreditation body’s own PT policy, or both where a laboratory holds more than one form of recognition).
- A procedure for handling PT samples that treats them identically to routine patient/client specimens — same analyst rotation, same instrument, no special repeat testing that wouldn’t happen on a real sample.
- A documented review of every PT result, not only the unsatisfactory ones, with a defined trigger for investigation (e.g., any warning-zone z-score, any unsatisfactory result, or a trend across multiple rounds).
- A corrective-action process that can be evidenced to an assessor or CMS surveyor on request, including root cause, corrective action taken, and verification that the action worked.
- Awareness that PT referral — sending a PT sample to another laboratory for analysis or consultation — is treated as a serious violation under CLIA in its own right, independent of how the underlying analyte performs.
Frequently asked questions
Is proficiency testing the same thing as external quality assessment (EQA)?
Functionally, yes. ISO/IEC 17043 covers both under a single framework; “external quality assessment” is simply the term many medical/clinical laboratory PT schemes use for the same activity of analyzing externally supplied, unknown-value samples and having performance evaluated by an outside provider.
How often does a laboratory have to do PT?
It depends on the specific requirement. Under CLIA (42 CFR Part 493, Subpart H), regulated analytes generally require a fixed number of testing events per year, with specific sample sets per event. Accreditation bodies operating under ISO/IEC 17025 set their own PT/participation frequency requirements as part of the laboratory’s accreditation scope, and these can vary by test and by accreditation body.
What is a z-score in proficiency testing, in plain terms?
It is a standardized way of expressing how far a laboratory’s reported result is from the reference value, in units of the scheme’s target standard deviation. A z-score near zero means the result was close to the reference value; conventionally, |z| ≤ 2 is treated as satisfactory, and |z| ≥ 3 as unsatisfactory, with the middle band flagged for review.
Can a laboratory send a PT sample to a reference lab or another site for help?
No. Under CLIA, this is PT referral, and it is prohibited and independently sanctionable — the whole point of PT is to evaluate the laboratory’s own routine testing process, not its ability to source a correct answer elsewhere.
Does failing one PT event mean a laboratory loses its CLIA certificate?
No, not on its own. CLIA’s regulatory definition of “unsuccessful performance” requires failing the same analyte in two consecutive testing events, or two of three consecutive events, before the cease-testing and corrective-action requirements are triggered. A single failed event still warrants investigation as good laboratory practice, but it is not itself a regulatory failure under CLIA’s definition.
For the broader accreditation context this obligation sits inside, see our guide to ISO/IEC 17025 and the Laboratory Compliance hub for related topics in accreditation, calibration, and quality systems.
