Lot to Lot Verification: Formula, Criteria, Worked Example

Lot to lot verification is the check a laboratory makes when a new lot of reagent (or calibrator) arrives: run the same patient samples on the old lot and the new lot, compare the results, and put the new lot into use only if the difference is small enough not to matter clinically. It takes an hour or two of bench time, and it is the difference between noticing a shift in creatinine on the day the lot changes and noticing it weeks later in a run of unexplained eGFR changes. This guide explains when it is needed, how many samples to use, the formula, how to choose acceptance criteria, two worked examples and a template you can copy.

Starter content. The procedure below is a practical summary for small and mid-sized labs. Your SOP, approved by your pathologist or biochemist, sets the actual sample numbers and limits; the CLSI EP26 guideline is the detailed reference many labs use.

What is lot to lot verification and when is it needed?

Manufacturers make reagents in lots (batches). Each lot is released within the manufacturer's specifications, but two lots of the same reagent can still differ slightly, and for some analytes and methods, particularly immunoassays, the difference can be large enough to change a patient's result category. ISO 15189 expects a lab to verify reagents before use and when a new lot or shipment could change performance. In plain words: before patient results are reported on a new lot, show that it agrees with the lot you trust.

Verify when:

• a new reagent lot arrives for an analyte;
• a new calibrator lot is introduced (often together with the reagent);
• the same lot arrives in a shipment that may have been stored or transported badly;
• the manufacturer changes the formulation, even under the same catalogue number.

Why not just run QC on the new lot?

Because QC material does not always behave like patient samples. Control material is a processed, stabilised matrix; a change in the reagent can shift the control and the patients differently, or shift one and not the other. This is called a lack of commutability. A review of lot-to-lot variance in immunoassays (PMC10252387) describes cases where internal QC did not reveal a clinically significant patient shift, and the reverse. Running QC on the new lot is still useful, but the lot decision should rest on patient samples run on both lots. Note also that QC targets may need to be re-established on the new lot; that is a separate task from lot verification.

How many samples, and at which levels?

A practical range is 5 to 20 patient samples per analyte. Fewer than five gives too little information; many routine chemistry analytes are verified with about 10. Use more for analytes where lots are known to vary, such as many immunoassays and tumour markers, and where a small shift changes a clinical decision.

• Spread the samples across the reportable range: low, normal and high values.
• Include medical decision levels, for example glucose near 126 mg/dL or creatinine near the upper reference limit.
• Use fresh patient samples that are stable for the time both runs take, and run old and new lot the same day, on the same analyzer, by the same staff where possible.
• Your SOP decides the number; write it down so every verification follows the same rule.

The lot to lot verification formula

For each sample i, with result on the old lot O_i and on the new lot N_i:

• Difference d_i = N_i − O_i
• Percentage difference %d_i = (N_i − O_i) ÷ O_i × 100
• Mean bias (%) = the average of the percentage differences
• SD of the differences (%) = the standard deviation of the percentage differences (with n − 1)
• % within limit = number of samples with |%d_i| ≤ limit ÷ number of samples × 100

Some labs use absolute differences for low concentrations, where a percentage exaggerates tiny changes (for example 0.1 mg/dL on a creatinine of 0.5 is 20 %). If you do, state in the SOP below which concentration the absolute limit applies.

Acceptance criteria: how to choose them

Decide the criteria before you run the samples, not after seeing the results. A common structure has three parts:

1. A per-sample limit. How large may a single difference be? Options labs use include a fraction of the total allowable error (TEa) the lab has chosen for the analyte, for example TEa ÷ 2 as a starting point; a critical difference based on biological variation; or the manufacturer's lot-to-lot claim. The choice and its source go into the SOP.
2. A minimum share of samples within the limit, for example 100 % or at least 9 of 10.
3. An optional mean-bias limit, so that many small differences in the same direction (a systematic shift) are caught even when each sample passes individually.

If the new lot fails, do not use it for patient results: hold it, repeat the comparison if a handling error is possible, inform the supplier, and keep using the old lot while it lasts. If the bias is consistent and clinically important, your pathologist may decide on other measures, but that is a clinical decision, not a software one.

Worked example: glucose, 10 samples

Illustrative example. Old lot GL-24-331, new lot GL-26-702, same analyzer, same day. Acceptance rule: every sample within ±5 %.

Result: mean bias = (1.39 − 1.14 + 2.11 + 1.92 − 0.85 + 2.11 + 1.14 + 1.86 + 1.12 + 2.05) ÷ 10 = +1.17 %; SD of the differences = 1.20 %; 10 of 10 within ±5 % (100 %). The new lot meets the criteria: accept.

Record card (sample data): Glucose · lot GL-24-331 → GL-26-702 · 10 samples · mean bias +1.17 % · accepted by the pathologist on 02 Nov 2026.

Worked example 2: creatinine, rejected

Illustrative example. Old lot CR-25-118, new lot CR-26-044. Acceptance rule: every sample within ±5 % and mean bias within ±5 %.

Result: mean bias +7.8 % (SD 0.73 %), 0 of 10 samples within ±5 %. Every difference is positive: this is a systematic shift, not random noise. Reject. Next steps: label the new lot "not for use", inform the supplier with the data, keep reporting on the old lot, and repeat the verification on a replacement lot. A creatinine shift of this size would also move eGFR values, so the pathologist may want to review any results already reported on the new lot.

Is there a lot to lot verification calculator?

You do not need one: the formulas above work in any spreadsheet. Put old and new results in two columns, % difference in a third as =(new−old)/old*100, then AVERAGE and STDEV of that column, and COUNTIF for the number within your limit. The risk with spreadsheets is not the arithmetic but the record: who ran it, which lots, which analyzer, and who decided. PathLab Software does the same arithmetic and keeps that record (see below).

Lot to lot verification template

Columns to copy into your form or spreadsheet:

What to record for the assessor

An assessor reviewing reagent verification usually wants to see, for a sample of lot changes: which lots, on which analyzer, when, how many samples, the acceptance criteria decided in advance, the results, the decision, and who made it. The decision should come before the first patient result on the new lot. Keep failed verifications too; a rejected lot with a supplier complaint is good evidence that the process works. The NABL software checklist covers the other records in the same file.

How PathLab Software does it

In PathLab Software, the Lot verification tab on the QC page records a verification against the test parameter, the analyzer, the old lot and the new lot, picked from your reagent inventory or typed in. You set the per-sample limit (suggested as TEa ÷ 2 when a TEa is set for the analyte, always editable), the minimum share within the limit and an optional mean-bias limit, and enter 5 to 20 sample pairs. The software calculates each difference, the mean bias, the SD and the share within the limit, draws a difference plot and a new-vs-old scatter, and suggests accept or reject. A pathologist decides; accepting marks the linked reagent batch "verified", and drawing stock from a batch whose verification is still pending raises a non-blocking warning. The record prints for the quality file. It is part of the QC statistics module with Sigma metrics and measurement uncertainty, included in the Complete and Network plans (pricing). For the uncertainty side of the same QC data, see the measurement uncertainty guide.

Record lot verifications against the reagent batch and print the record: try PathLab Software free for 21 days.

Questions people ask

What is lot to lot verification in a medical laboratory?

Lot to lot verification is the comparison a medical laboratory makes before using a new reagent or calibrator lot: the same patient samples are run on the old and new lots, and the new lot is accepted only if the differences meet limits set in advance. It protects patients from result shifts caused by lot changes.

What is the formula for lot to lot verification?

For each sample, percentage difference = (new lot result − old lot result) ÷ old lot result × 100. The mean bias is the average of these percentages, the SD of the differences shows their spread, and the share within limit is the number of samples inside the per-sample limit divided by the total.

What are the acceptance criteria for lot to lot verification?

Labs usually set a per-sample limit, such as a fraction of the total allowable error they chose, a critical difference from biological variation or the manufacturer's claim; a minimum share of samples within it, often 100 %; and optionally a mean-bias limit. Criteria are decided and documented before the samples are run.

How many samples are needed for reagent lot verification?

A practical range is 5 to 20 patient samples, spread across the reportable range and including medical decision levels; about 10 is common for routine chemistry. Use more for analytes where lots vary more, such as many immunoassays. Your laboratory's SOP sets the number.

Is running QC on the new lot enough?

Usually not. QC material may not change the way patient samples do when a reagent lot changes, because of commutability differences, so a patient shift can be missed. Base the lot decision on patient samples run on both lots, and re-establish QC targets on the new lot as a separate step.

Is there a lot to lot verification calculator?

A spreadsheet with three formulas is enough: percentage difference per sample, the average and standard deviation of those differences, and a count of samples within your limit. PathLab Software calculates the same values, draws the difference plot and keeps the record of lots, analyzer and decision.