Which supplier and lot changes force full revalidation, and which only need a

A 40-vial COD run came back with a calibration curve RSD of 4.8% against a method limit of 2.0%, and every vial in the failing set traced to a single lot change: the supplier had moved from a 1.0 mm PTFE-faced silicone septum to a 1.3 mm laminate with a different silicone cure. Nothing in the method text changed. The certificate of analysis still read “PTFE/silicone, 8 mm.” The lab had no revalidation trigger written down, so the lot went straight onto the autosampler. That is the failure mode this article is about: not whether you validate a COD method, but how you decide, in writing, which supplier or lot changes force you back into validation and which only need a verification run.

If you already treat the cod vial as a method component, the next question is procedural. A change in that component is a change to the method. The trick is that not every change is equal, and the inspector who reads your change-control entry wants to see that you made the distinction on purpose, with data, before you released the affected lots.

Which supplier and lot changes force full revalidation, and which only need a verification run

The dividing line is whether the change can move a measured result outside the method’s stated performance. I use a three-tier system. It is not in any single regulation, but it maps cleanly onto the change categories in EPA SW-846 method modification guidance and onto USP <621> system suitability expectations, and it survives audit because each tier has a defined test and a defined acceptance criterion.

Tier 1: revalidation required

Any change that touches the wetted surface chemistry, the septum formulation, the septum thickness by more than 0.1 mm, the crimp geometry, or the glass hydrolytic class. These are the variables that change extractables, seal integrity, or headspace volume. A new septum compound can shed siloxane oligomers into the headspace and show up as a baseline shift on a COD digest read at 600 nm. A crimp geometry change alters the seal compression and therefore the vapor loss path. A glass class change from Type I borosilicate to Type II alters alkali leaching, which matters for any method with a long digestion at 150 °C.

Tier 2: verification only

Same design, same materials, same drawing revision, new manufacturing lot or new production line at the same supplier. You are confirming that the lot behaves like the validated design, not re-proving the design. A verification run is a system suitability injection set plus a spike recovery at one concentration. If the lot passes, it goes into the validated envelope and you document why.

Tier 3: no action beyond incoming inspection

Packaging changes, carton count, label format, color of the cap, a supplier’s internal ERP change that does not touch the drawing. These are not method changes. Log them, inspect to the AQL you set for the defect class, and move on.

The table is the decision rule. Write it into your change-control SOP as an appendix, not as prose buried in section 7.3. An inspector who can point at a table and then at the corresponding entry in your change log has nothing left to ask.

The dimensional stack is where most “small” changes actually bite

I have watched a Tier 2 change get downgraded to Tier 3 by a lab manager because the supplier swore the drawing had not moved. The drawing had not moved. The molding tool had. The new tool produced a septum seat 0.15 mm deeper, which pulled the punctured septum face 0.15 mm further from the needle guide. On a 2 mL autosampler vial that is enough to change where the needle pierces, and a needle that pierces at a different radius on the septum produces a different coring pattern. The dimensional stack is unforgiving: cap height, septum thickness, crimp depth, and neck finish all have to land inside the same envelope the method was validated against.

If your method uses a 2 mL vial format with a standard 11 mm crimp, the validated stack is typically: 32 × 11.6 mm body, 8 mm septum, 1.0 mm PTFE-faced silicone, crimped to 1.0-1.4 N·m. Any supplier change that moves a dimension inside that stack by more than the drawing tolerance on the critical dimension (septum thickness, seat depth, crimp skirt height) is a Tier 1 event in my book, regardless of what the supplier calls it. The reason is simple: the method was validated against a physical object, and the physical object changed.

How to measure the stack cheaply

You do not need a metrology lab. A digital micrometer with 0.001 mm resolution, a crimping tool with a calibrated torque setting, and a go/no-go gauge for the neck finish will catch 90% of stack drift. Measure 10 vials per lot: septum thickness, cap skirt height, and overall crimped height. Record the mean and range. If the mean shifts by more than 0.05 mm from the validated baseline, escalate to Tier 2 at minimum. If the range exceeds the drawing tolerance, reject the lot.

Writing the change-control entry so it survives inspection

Most change-control entries fail not because the science is wrong but because the entry does not show the reasoning. The inspector is looking for four things, in this order: what changed, why you classified it the way you did, what test you ran, and what the result was against a pre-defined criterion. If any of those four is missing, you get a 483 observation on change control even if your data is clean.

The four-part entry

• Change description: name the supplier, the lot, the drawing revision, and the specific attribute that moved. “Septum compound changed from Lot A (silicone, 50 Shore A) to Lot B (silicone, 55 Shore A) per supplier CoA dated 2024-03-11” is a description. “Septum changed” is not.
• Classification rationale: state the tier and the reason. “Classified Tier 1 because septum compound is a wetted-surface material that affects extractables and seal integrity.” Reference the SOP appendix.
• Test protocol: name the method, the number of replicates, the concentrations, the acceptance criteria, and the analyst. Pre-define the criteria before you run the test. An inspector will ask whether the criterion was written before or after the data.
• Result and disposition: the measured values, the pass/fail against the pre-defined criterion, and the release decision. If it failed, the entry must show the corrective action and the re-test.

The one thing I insist on in every entry is a reference to the validated baseline. If your validation package recorded a system suitability RSD of 0.8% on the original design, your verification run has to be compared to 0.8%, not to the method limit of 2.0%. A lot that comes in at 1.9% passes the method but has drifted 2.4× from baseline. That is a signal, and an inspector who sees you noticed it will trust the rest of the entry more.

The comparison table that belongs in the entry

Put a small table in the entry itself, not in an appendix. Columns: attribute, validated baseline, current lot, delta, criterion, pass/fail. Six rows: system suitability RSD, spike recovery, blank absorbance at the read wavelength, septum thickness, crimp torque, headspace volume. That table is the single most useful artifact in the whole package because it lets a reviewer see the drift in one glance.

Two field cases where the trigger was missed and what caught it

Case 1: the baseline shift that looked like a reagent problem

A wastewater lab running COD by the closed-reflux colorimetric method saw blank absorbance climb from 0.041 to 0.068 AU over two weeks. The analyst blamed the digestion reagent and ran three new reagent lots. The drift persisted. The measurement that identified the real cause was a septa-swap experiment: same reagent, same standards, vials from the old lot versus vials from the new lot, six replicates each. The new-lot blanks averaged 0.067 AU; the old-lot blanks averaged 0.043 AU. The difference was the septum. The new lot had a laminate with a thinner PTFE face, and the silicone underneath was off-gassing at the 150 °C digestion temperature. The fix was a Tier 1 revalidation with the original septum specification restored, plus a written rule that any septum face thickness change under 0.1 mm still requires a blank absorbance check at the read wavelength before release.

Case 2: the recovery drop that only showed at the low end

A contract lab running COD on industrial effluent saw spike recovery at 50 mg/L fall from 96% to 84% after a supplier moved production to a second line. Recovery at 500 mg/L was unchanged at 98%. The measurement that caught it was a five-point recovery curve rather than the single mid-point spike the SOP called for. The low-end drop traced to a crimp geometry change: the new line’s crimper produced a 0.12 mm shorter skirt, which reduced seal compression and allowed a small vapor loss during the 2-hour digestion. At high concentration the loss was inside the noise; at low concentration it was a 12-point recovery miss. The fix was to add a low-end spike (at 20% of the calibration range) to the Tier 2 verification protocol and to reject the affected lots. That single protocol change has caught three more supplier drifts since.

Both cases share a pattern: the trigger was real, the change was documented by the supplier as minor, and the only thing that caught it was a test designed to detect drift rather than confirm compliance. That is the difference between a verification run and a compliance check.

What to put in the revalidation protocol when Tier 1 is triggered

A Tier 1 event means you re-establish the method’s performance characteristics against the new component. For a COD method that is a defined set of experiments, and you should not let anyone talk you into a shorter list.

• Calibration: minimum five points across the working range, three replicates each, RSD ≤ 2.0% at each level, r² ≥ 0.999.
• Accuracy: spike recovery at 20%, 50%, and 100% of the range, n=3 each, acceptance 90-110% (some methods allow 80-120%; use whatever the method text specifies and do not widen it).
• Precision: repeatability n=6 at mid-range, RSD ≤ 2.0%; intermediate precision across two analysts and two days, RSD ≤ 3.0%.
• LOD/LOQ: re-establish from the standard deviation of seven low-level replicates; LOQ typically 3.3× LOD.
• Blank: absorbance at the read wavelength ≤ 0.050 AU for the new component, compared to the validated baseline.
• Carryover: blank after a high standard ≤ LOQ.
• Robustness: one deliberate variation (digestion time ± 10 min, or temperature ± 2 °C) to confirm the method still holds with the new component.

If the component is a vial insert or a 2 mL format with a micro-insert, add a headspace volume check: the insert reduces the available headspace by a known amount, and the validated method was tuned to that volume. A change in insert wall thickness of 0.2 mm can shift the headspace by 5%, which is enough to move a gas-phase equilibrium.

The supplier-side evidence that lets you downgrade a trigger

You can downgrade a Tier 1 to a Tier 2 only if the supplier gives you evidence that the change does not touch the critical attributes. In practice that means three documents: a drawing revision comparison showing the critical dimensions unchanged, a material certification showing the same compound and the same cure, and a capability study (Cpk) on the critical dimensions from the new line or lot. If the supplier cannot produce all three, the change stays Tier 1. I have never regretted holding that line.

The capability study is the one most suppliers cannot produce, and it is the most useful. A Cpk ≥ 1.33 on septum thickness and crimp skirt height tells you the new line is in control. A Cpk of 0.9 tells you the line is producing parts that will drift into your method. If you are sourcing through a distributor, ask for the manufacturer’s capability data, not the distributor’s incoming inspection. The HPLC consumables supply chain is long, and the data quality drops with every handoff.

A note on the standards you can point at

There is no single clause in SW-846 that says “revalidate when the vial changes.” What the methods do say is that the analyst is responsible for demonstrating that the method performs as intended for the sample matrix and the equipment used. That is the hook. Your change-control SOP is the mechanism that shows you took that responsibility seriously. For the chromatography side, the system suitability sections in USP <621> and the equivalent in your method text give you the acceptance criteria. For consumable specifications, the manufacturers publish the dimensional and material data you need to write a defensible drawing: Agilent’s consumables documentation and Thermo Fisher’s sample handling specifications are both good models for what a supplier should be giving you. Restek’s technical resources are worth reading for the septa and inlet consumables background that translates directly to COD vial closures.

The decision rule to write into your SOP

Classify every supplier or lot change against the three tiers before the material reaches the bench. Tier 1 (wetted surface, septum compound or thickness, crimp geometry, glass class) forces full revalidation with the protocol above. Tier 2 (new lot, new line, same drawing) needs a verification run with system suitability plus a low-end spike recovery at 20% of the range. Tier 3 (packaging, labeling) needs incoming inspection only. Write the four-part change-control entry for every Tier 1 and Tier 2 event, pre-define the acceptance criteria, and compare against the validated baseline rather than the method limit. If the supplier cannot produce a drawing comparison, a material certification, and a Cpk study on the critical dimensions, the change stays Tier 1 regardless of what the change notice says. That rule has kept more calibration curves inside 2.0% RSD than any amount of retraining.

References & further reading

Related resources from our lab consumables network:

• Cod Vial at codvial.com
• Vial Insert at hplcinsert.com
• 2Ml Vial at 2mlsamplevial.com
• Hplc Consumables at hplcconsumable.com

External references and standards:

• Agilent chromatography consumables and columns
• Thermo Fisher chromatography and sample handling
• U.S. EPA SW-846 test methods for waste and materials
• Restek technical resources for chromatography