Reagent Lot Verification in Clinical Biochemistry: A Practical Guide

Author's Note: This article is intended for clinical laboratory scientists, quality officers, and laboratory managers involved in routine chemistry, immunoassay, haematology, or point-of-care testing. It covers the rationale, standard methodology, real-world variation, and failure management for new reagent lot verification.

What Is Reagent Lot Verification — and Why Does It Matter?

Every time a clinical biochemistry laboratory receives a new lot (batch) of reagents — whether for glucose, troponin, creatinine, thyroid function, or any other analyte — a fundamental question arises: does this new lot perform the same as the old one?

Reagent lot verification is the formal process of answering that question before the new lot goes into routine patient use. It is not simply a box-ticking exercise. It is one of the most clinically significant quality practices a laboratory performs, because undetected lot-to-lot reagent shifts can silently move patient results by enough to alter clinical decisions.

Consider this scenario: a patient with chronic kidney disease has a stable creatinine of 145 µmol/L on the previous reagent lot. A new lot is introduced without verification. Due to a manufacturing shift, the new lot consistently reads 8% lower. The patient's next result comes back as 133 µmol/L — a change the clinician interprets as genuine improvement in renal function. The dose of a nephrotoxic drug is maintained rather than reduced. The patient is harmed — not by a machine error or a clinician error, but by an invisible reagent shift that was never caught.

This is not a hypothetical. Lot-to-lot shifts are a well-documented phenomenon in clinical chemistry, and the literature is clear: reagent manufacturers produce lots within stated specifications, but those specifications allow for variation. It is the laboratory's job to determine whether that variation is acceptable for its patient population and clinical context.

The Regulatory and Accreditation Framework

Reagent lot verification is mandated or strongly recommended by all major accreditation bodies and regulatory frameworks:

ISO 15189:2022 (the international standard for medical laboratory quality) requires laboratories to verify that new reagent lots perform equivalently to the previous lot before use. It specifically requires that the verification is documented and that acceptance criteria are defined in advance.

NATA/RCPA (Australia) require evidence of lot verification as part of quality system documentation. Laboratories accredited under ISO 15189 via NATA are expected to have a written procedure and records for every lot change.

CAP (College of American Pathologists) requires verification of new reagent lots and specifies that laboratories must compare results to established reference intervals and performance specifications.

CLIA (US) does not prescribe the exact method but requires that laboratories establish performance specifications — including accuracy and precision — for each method, and that these specifications are maintained when reagents change.

CLSI EP26-A (User Evaluation of Between-Reagent-Lot Variation) is the primary technical guidance document. It provides statistical frameworks for designing lot comparison studies and interpreting results.

Despite this framework, the specific protocols vary considerably between laboratories, and that variation is where most of the practical challenge lies.

Standard Practice: What Most Laboratories Do

The majority of accredited clinical biochemistry laboratories follow a broadly similar approach, anchored in a comparison between the outgoing lot and the incoming lot.

The Core Approach

A set of patient or control samples is measured using both the current (outgoing) lot and the new (incoming) lot, ideally on the same analyser in close temporal succession. The results are compared using predefined acceptance criteria. If the new lot performs within acceptable limits, it is released for routine use. If not, it is investigated before patient use proceeds.

Sample Selection

Most laboratories use a combination of:

• Internal quality control (IQC) materials — commercially prepared materials at two or three concentration levels (typically low, medium, and high). These are convenient, readily available, and provide a direct comparison point.
• Residual patient samples — leftover patient specimens from recent routine runs. These are particularly valuable because they represent the true matrix in which the analyser will operate. They also provide clinical range coverage that may not be achieved with QC materials alone.

The number of samples varies. A pragmatic minimum used by many laboratories is 20 paired samples (same sample on both lots). CLSI EP26-A recommends a minimum of 40 pairs for robust statistical conclusions, though many busy clinical labs operate with fewer due to practical constraints.

Analytes Included

Not all analytes are verified with the same rigour. Most laboratories risk-stratify their verification:

• High clinical risk analytes (troponin, creatinine, sodium, potassium, glucose, coagulation tests) receive the most rigorous verification, often including patient samples across the clinical decision range.
• Moderate risk analytes (ALT, bilirubin, albumin) typically receive comparison with QC materials plus a subset of patient samples.
• Lower risk or rarely changed reagents may be verified with QC materials only, provided the manufacturer's lot-to-lot specifications are reviewed.

How It Varies from Lab to Lab

Despite the common framework, verification practice differs substantially across institutions. Understanding this variation helps laboratories benchmark their own approach and identify gaps.

Variation in Sample Numbers

A large tertiary hospital laboratory processing thousands of samples per day may run 40 patient pairs plus three QC levels across six lots on an analyser platform. A small regional hospital might run five patient samples and two QC levels, accepting the manufacturer's specifications as the primary safeguard.

Neither approach is inherently wrong — the appropriateness depends on clinical risk, volume, and the laboratory's stated acceptance criteria. The problem arises when sample numbers are insufficient to detect a clinically meaningful shift with adequate statistical power.

Variation in Acceptance Criteria

This is where laboratories diverge most significantly. Acceptance criteria define how large a difference between the old and new lot is tolerable before rejecting the new lot.

Common approaches include:

Total Allowable Error (TEa)-based criteria: The bias between lots must be less than a defined fraction of the TEa for that analyte. For example, if the TEa for creatinine is 10%, and the laboratory uses 50% of TEa as its acceptance threshold, then the new lot must agree with the old lot to within 5%. This approach links acceptance to clinical outcome.

Bias percentage: A simpler approach in which the mean bias between lots must fall within a fixed percentage, often derived from the manufacturer's stated lot-to-lot variability. Common thresholds are ±5% for most chemistry analytes, ±10% for immunoassays.

Reference interval shift: The laboratory checks whether the new lot would produce results that shift significantly within or across clinically important decision limits (e.g., does the eGFR calculation still correctly classify patients with CKD Stage 3?).

Statistical significance (p-value): Some laboratories rely on a paired t-test, rejecting a lot if p < 0.05. This approach is widely discouraged in current guidance because statistical significance does not equal clinical significance — a very large study might detect a 0.5% bias that is of no clinical consequence, while a small study might miss a 6% bias that is highly consequential.

Variation in Timing

Some laboratories complete lot verification before opening the new lot box at all, running all comparison testing in advance. Others open the new lot simultaneously and run it in parallel for one to three days before switching. A minority switch immediately and rely on IQC performance retrospectively — an approach that is difficult to defend under ISO 15189 scrutiny.

Variation in Who Performs and Authorises Verification

In larger laboratories, verification may be performed by a dedicated quality team with sign-off by the laboratory medical director. In smaller labs, a senior scientist performs and authorises verification independently. Accreditation bodies expect documented authorisation, but the level of seniority required varies between institutions.

The Recommended Method: A Best-Practice Framework

Based on CLSI EP26-A, ISO 15189 requirements, and established laboratory medicine principles, the following approach represents current best practice for most clinical biochemistry settings.

Step 1: Define Acceptance Criteria Before You Start

This is the most commonly skipped step — and the most important. Acceptance criteria must be set before any data are collected, not after. Setting them post-hoc introduces unconscious bias and is indefensible under audit.

For each analyte subject to verification, document:

• The TEa (from RCPA or equivalent national scheme allowable limits, or CLIA proficiency criteria)
• The acceptance threshold (recommended: bias ≤ 1/3 to 1/2 of TEa)
• The number of samples to be tested
• The concentration range to be covered
• Who will review and authorise results

Example — Creatinine (enzymatic method, Roche cobas):

• TEa: 10% (RCPA allowable limit)
• Acceptance criterion: Mean bias ≤ 5% at all concentration levels
• Samples: 20 patient samples spanning 60–600 µmol/L; 3 QC levels (low ~80, mid ~200, high ~500 µmol/L)
• Authorisation: Senior Scientist + Laboratory Manager

Step 2: Select and Prepare Samples

Collect residual patient samples that have been stored appropriately (most chemistry analytes: 2–8°C for up to 48 hours; frozen aliquots for immunoassays). Supplement with commercial QC materials at low, medium, and high concentrations.

Exclude samples that are:

• Haemolysed, lipaemic, or icteric beyond the interference thresholds for the analyte
• Outside the analytical measuring range of the method
• More than 24–48 hours old (unless frozen at the time of collection)

Ideally, samples should span the full clinically relevant range, including concentrations around key decision points. For creatinine, ensure samples near the upper reference interval (approximately 110 µmol/L in women, 130 µmol/L in men), the CKD Stage 3 threshold (around 150–180 µmol/L range for eGFR ~45), and the renal failure range (>400 µmol/L).

Step 3: Run the Comparison

Run each sample on the current lot (Lot A) and the new lot (Lot B) within the shortest feasible time window — ideally within two hours to minimise analyte degradation effects. If the analyser allows, alternate runs between lots to distribute any within-day drift evenly.

Each sample should be run in duplicate if resources allow; at minimum, duplicate QC measurements on each lot.

Record the lot numbers, expiry dates, calibration status of each lot, and the run times.

Step 4: Calculate and Evaluate Bias

For each analyte, calculate the following:

Mean bias (absolute and percentage):

Bias% = [(New Lot Mean − Old Lot Mean) / Old Lot Mean] × 100

Do this separately for each concentration range (low, medium, high), not just overall, because bias is often concentration-dependent.

Regression analysis (for larger sample sets): A Passing-Bablok or Deming regression on the paired patient results will reveal whether any bias is constant (additive) or proportional. This matters because a 5% proportional bias at high concentrations may be clinically unacceptable even if the low-concentration results are fine.

QC comparison: Check whether the new lot IQC results fall within the existing lot's IQC mean ± 2SD. If they do not, this is a signal of a lot-dependent shift, even before patient data are evaluated.

Step 5: Apply Acceptance Criteria

Compare the calculated bias at each concentration level against the pre-defined acceptance criteria. Document the comparison explicitly.

If all criteria are met: approve the new lot and document the approval.

If one or more criteria fail: do not proceed to routine use until the failure is investigated and resolved (see below).

Step 6: Document and Archive

The verification record must include:

• Lot numbers (old and new), expiry dates
• Reagent and calibrator positions used
• QC data for both lots
• Patient sample identifiers and results (both lots)
• Calculated bias and regression statistics
• Acceptance criteria (pre-defined)
• Pass/fail outcome
• Authorising signature and date
• Any corrective actions taken (if failure occurred)

This documentation must be retained for the period required by your accreditor — typically a minimum of five years under ISO 15189.

Realistic Worked Example: Sodium on an Integrated Chemistry Analyser

Setting: A 400-bed public hospital laboratory. Platform: Beckman Coulter AU5800. Analyte: Sodium (indirect ISE). Lot change: Current lot (Lot 2211A) has 3 weeks remaining; new lot (Lot 2307B) has arrived.

Pre-defined criteria (documented in SOP):

• TEa for sodium: ±4 mmol/L (RCPA allowable limit)
• Acceptance criterion: Mean bias ≤ 2 mmol/L at all levels
• Samples: 20 residual patient samples (concentrations ranging from 128 to 152 mmol/L) + QC low (125 mmol/L target), QC normal (140 mmol/L target), QC high (155 mmol/L target)
• Clinically critical decision points: 130 mmol/L (hyponatraemia treatment threshold), 145 mmol/L (upper reference limit)

Results:

Interpretation:

Mean bias across all levels: +0.7 mmol/L (0.5%). This is well within the acceptance criterion of ≤ 2 mmol/L. The bias is constant (additive), which is consistent with a minor calibration offset between lots. No concentration-dependent bias is observed. Regression analysis on patient pairs shows a slope of 1.002 and intercept of −0.06 — essentially no proportional bias.

Decision: Lot 2307B is approved for routine use. The existing QC target means are updated to reflect the new lot values (+0.7 mmol/L across levels), and the IQC charts are annotated with the lot change date.

Clinical significance check: Would any patient near 130 mmol/L be misclassified? A patient with a true sodium of 130 mmol/L would be measured as ~131 mmol/L on the new lot — still firmly in the hyponatraemic range. No clinical impact.

Realistic Worked Example: TSH Immunoassay Fails Verification

Setting: Same hospital. Platform: Roche cobas e801. Analyte: TSH (third-generation immunoassay). Lot change: Current lot (Lot 19384A) nearing expiry; new lot (Lot 19421C) received.

Pre-defined criteria:

• TEa for TSH: ±20% (immunoassays carry higher inherent variability)
• Acceptance criterion: Mean bias ≤ 10% across all concentration ranges
• Samples: 25 residual patient samples (spanning 0.08–45 mIU/L) + 3 QC levels
• Critical decision point: 0.1 mIU/L (suppressed TSH threshold for thyrotoxicosis)

Results:

Interpretation:

At mid and high concentrations, the new lot performs equivalently. However, at low concentrations — clinically critical for detecting suppressed TSH and monitoring hyperthyroidism treatment — the new lot shows a +24% positive bias that exceeds the 10% acceptance criterion.

This matters enormously: a patient with true TSH of 0.08 mIU/L (suppressed, consistent with hyperthyroidism) would be measured as 0.10 mIU/L on the new lot — crossing the "suppressed" threshold and potentially being reclassified.

Decision: Lot 19421C fails verification. It is quarantined. Routine use does not proceed.

What to Do When Verification Fails

A failed lot verification is not a crisis — it is the quality system working exactly as intended. The laboratory has detected a problem before patients are affected. The response should be systematic and documented.

Immediate Steps

1. Quarantine the new lot. Physically segregate the failing lot with a clear "Do Not Use" label. Ensure all staff are notified that the lot switch is on hold.

2. Document the failure. Record the specific acceptance criterion that was not met, the magnitude of the failure, and the date of discovery. This becomes part of the lot verification record and feeds into any subsequent non-conformance report.

3. Continue with the current lot. If the outgoing lot has sufficient volume remaining, continue patient testing on it while the investigation proceeds. If the outgoing lot is depleted or expired, escalate immediately (see below).

Investigation Checklist

Work through these possibilities in order:

Was the comparison conducted correctly?

• Were samples run correctly on both lots?
• Was calibration completed on the new lot before comparison testing?
• Were the samples within their stability window?
• Was the same QC material used for both lots, and was it in date?

Errors at this step are common, particularly when verification is performed under time pressure. Repeat the verification before assuming the lot itself is at fault.

Is the failure specific to one concentration range?

As in the TSH example above, a lot may fail at low concentrations but pass at mid and high ranges. This is worth investigating further with the manufacturer, who may be able to advise on lot-specific performance characteristics or calibrator adjustments.

Is the analyser involved?

Run the new lot on a second analyser (if available) to determine whether the bias is lot-specific or instrument-specific. An instrument-related problem (dirty sample probe, degraded ISE membrane, drift in photometric path) can masquerade as a lot failure.

Contact the manufacturer's technical support.

Provide them with the lot number, verification data, and the specific failure details. Manufacturers track lot performance across customer sites, and they may be aware of a production issue with this lot. They may offer a replacement lot, a revised calibration protocol, or performance documentation to support conditional acceptance.

Ask for the manufacturer's stated lot-to-lot variability data (Certificate of Analysis) for both lots. If their own data shows the new lot is at the edge of specification, that supports a genuine lot problem.

Conditional Acceptance (with Recalibration)

In some cases, the manufacturer may advise a specific calibration adjustment for the new lot. If this resolves the bias within acceptance criteria, the lot may be accepted conditionally, with the adjustment documented and its effect verified with a second comparison run.

This is acceptable under ISO 15189 provided the acceptance criteria are still met post-adjustment and the adjusted calibration is maintained.

Obtaining a Replacement Lot

If investigation confirms a genuine lot problem, request a replacement lot from the manufacturer or distributor. This is standard procedure and reputable manufacturers will process urgent replacements for in-vitro diagnostic products.

In the interim, consider:

• Send-away referral: For analytes where the lot failure precludes testing, arrange urgent send-out to a reference laboratory.
• Alternative analyser: If a backup analyser is available with a verified reagent lot, redirect urgent samples.
• Clinical notification: If there is any risk that patient results may already have been affected (e.g., if the new lot was inadvertently used before the failure was detected), notify clinical teams and consider result review.

When the Outgoing Lot Is Expired or Depleted

This is the most challenging scenario. If the current lot has run out and the new lot has failed verification, the laboratory must:

• Notify laboratory management and the medical director immediately.
• If a replacement lot can be obtained within hours, extend the outgoing lot use by exception with documented clinical justification (acceptable for a short period in most accreditation frameworks, with explicit risk assessment).
• If testing cannot continue safely, suspend the test and redirect patients to an alternative testing site.
• Document every decision and the clinical rationale.

Practical Tips for a Smooth Verification Process

Stagger lot changeovers. Avoid changing multiple reagent lots simultaneously — diagnosing the source of a problem becomes impossible if three lots have changed at once.

Build in a buffer. Initiate verification when the current lot still has at least 10–15% volume remaining. This provides time for repeat testing if the first verification fails.

Keep verification samples. After a successful verification, retain the patient samples used (frozen where possible) as a reference set. If a QC problem arises weeks later, you can use these to determine retrospectively whether a subsequent lot shift was the cause.

Automate the calculations. Create a standardised spreadsheet or LIS-based template that calculates bias, applies acceptance criteria, and generates a pass/fail output automatically. Manual calculations introduce transcription errors and are time-consuming.

Review IQC trends post-switchover. Even after a successful verification, monitor the first two to three days of IQC data on the new lot for unexpected drift. Verification studies provide a point-in-time snapshot; IQC provides ongoing surveillance.

Include verification in your lot tracking register. Maintain a rolling register of all lot changes, verification dates, outcomes, and authorising scientists. This register is a key document during accreditation audits.

Summary

Reagent lot verification is not a regulatory formality — it is an act of patient safety. The fundamental principle is simple: assume nothing, verify everything, and document the evidence. A robust verification programme detects clinically meaningful lot-to-lot shifts before they affect patients, protects the laboratory under audit, and builds clinician confidence in the results the laboratory produces.

The investment of time in defining clear acceptance criteria, selecting appropriate samples, running a structured comparison, and managing failures systematically pays dividends every time a reagent lot shifts in a way that would otherwise go unnoticed — because in clinical biochemistry, what you don't measure, you don't know.