SOP: CDS Manual Integration, Reprocessing, and Testing-Into-Compliance Prevention

This is a ready-to-use SOP. Replace every <<FILL: ...>> placeholder with your own specifics, set your document numbers and dates, and route it through your normal document control, review, and approval. It is written to be the single procedure that governs processing method locking, manual integration, reprocessing review, and escalation for a chromatography data system, rather than leaving each of those controls to live in a separate, loosely connected document. A worked filled specimen follows the template. This is general guidance to adapt and verify against your own regulatory and quality system requirements, not legal or regulatory advice.

Document control header

1. Purpose

This procedure defines how <<FILL: COMPANY NAME>> controls chromatographic integration on a chromatography data system (CDS) so that the reported result is the scientifically justified one, not merely the one that happened to pass. It governs when the locked, validated processing method is the only integration used, when and how manual integration may be performed instead, how reprocessing events are reviewed, and when an unfavorable result must be escalated to an out-of-specification (OOS) or deviation investigation rather than reprocessed, retested, or quietly discarded.

2. Scope

This procedure applies to all GMP chromatographic testing performed on a CDS at the sites listed in the header, covering high performance liquid chromatography, gas chromatography, and any other technique run on the same class of software. It covers the processing method itself, every manual integration, every reprocessing event, and the point at which an unfavorable result must leave this procedure and enter OOS or deviation handling. It does not replace the second-person review procedure, <<FILL: SOP-ID for second-person review of laboratory data>>, which this procedure feeds, and it does not replace the audit trail review procedure, <<FILL: SOP-ID for audit trail review>>, which this procedure relies on for the mechanics of reading the trail.

3. Responsibilities

4. Definitions

• Locked processing method: a version-controlled set of integration parameters (threshold, peak width, baseline mode, inhibit windows) developed and validated with the analytical method, applied by default to every injection, and changeable only through change control, not by an analyst at the bench.
• Manual integration: a per-injection override of the locked method’s automatic result, where the analyst redraws a baseline or sets a peak boundary by hand.
• Reprocessing: any event where a result set is regenerated from the acquired signal, whether using the locked method, a manual override, or a corrected calibration or sequence parameter, producing a new versioned result.
• Testing into compliance: generating a result, evaluating it outside the official record or reprocessing it repeatedly, and reporting only a favorable version without the investigation an unfavorable result requires. See 21 CFR 211.192 and 211.194(a) and the FDA 2018 data integrity guidance.
• Result-driven change: any reprocessing or manual integration whose only discernible justification is that it produced a passing value, as opposed to a scientific reason recorded before the outcome was known to be favorable.

5. Procedure

5.1 Lock the processing method and control changes to it

1. Every validated analytical method has one associated, version-controlled processing method in the CDS, applied automatically to every injection acquired against it.
2. The processing method is not editable by analysts. Changes are made only by the method owner, under change control, and are themselves audit-trailed.
3. The locked method’s default output is what section 5.4 evaluates first, before any manual integration is considered.

5.2 Define when manual integration is permitted

1. Manual integration is permitted only for a documented, foreseeable integration limitation of the locked method: a known co-eluting shoulder, a rider peak, or a baseline disturbance the method appendix already describes, or a genuinely novel scientific circumstance approved by the method owner or a designated technical reviewer before the result is reported.
2. Manual integration is never permitted solely because the locked method’s result fails or is borderline. A failing or borderline locked-method result is data; it is handled per section 5.5, not overridden to avoid it.
3. Manual integration is flagged in the CDS audit trail by configuration; if the system requires a setting to make manual events distinguishable from automatic ones, that setting is verified as part of CDS validation, not assumed.

5.3 Document and review every manual integration

1. The analyst completes a manual integration justification record for every manual integration, before the result is reported. Use the manual integration justification form: sample and run ID, peak affected, the specific reason, the automatic (before) and manual (after) result, whether the change crosses a specification limit, and the scientific rationale.
2. A second-person reviewer, independent of the analyst, reviews the record and the CDS audit trail entry together, confirms the reason is specific and method-based rather than a phrase like “improved integration,” and confirms the manual result is not merely the passing one when the automatic result was not.
3. Any manual integration that moves a result across a specification limit requires a stronger, explicitly documented justification and is flagged to the laboratory supervisor at the time of review, not discovered later.
4. The reviewer’s decision (accept or reject the manual integration) is recorded on the form. A rejected manual integration means the locked-method result stands and is handled per section 5.5 if it is unfavorable.

5.4 Review the reprocessing chain, not just the final version

1. At second-person review, the reviewer opens the CDS result audit trail for the run, not only the printed report, and examines every version in the reprocessing chain: who reprocessed, when, what changed, and the result at each version. Use the task-level steps in the CDS audit trail review work instruction for how to read the trail itself.
2. Where a reprocessing chain shows a result migrating toward or across a specification limit across successive versions, the reviewer treats this as a signal requiring explanation, not a routine event. A migration with no documented scientific basis for each step is a data integrity concern per section 5.6, not a result to accept because the final version passed.
3. Reconcile the injection sequence for the run, using the injection sequence reconciliation worksheet, so that no reprocessing or manual integration is hiding behind an unexplained orphan injection.

5.5 Escalate unfavorable results, do not reprocess around them

1. When the locked processing method produces a result that fails or is borderline against the specification, and no documented, pre-existing scientific basis justifies a manual integration under section 5.2, the analyst does not reprocess further to search for a passing value.
2. The unfavorable locked-method result is reported and routed to OOS or deviation handling per <<FILL: SOP-ID for OOS investigation>> or <<FILL: SOP-ID for deviation management>> as applicable.
3. A system suitability failure encountered during the same run is handled the same way: retained, investigated, and not simply rerun until it passes. See <<FILL: SOP-ID or cross-reference for system suitability handling>>.
4. Reprocessing a result that has already been escalated to OOS is permitted only as part of the OOS investigation itself, under that procedure’s controls, not as a parallel route back to a passing value.

5.6 Trend and periodically review manual integration and reprocessing

1. Laboratory supervision compiles, at least <<FILL: monthly>>, the count and rate of manual integrations and multi-version reprocessing chains by analyst and by method.
2. An analyst or method with a manual integration or reprocessing rate materially above the laboratory’s historical baseline is reviewed specifically: is the method poorly suited to routine automatic integration (a method design issue) or is the pattern analyst-specific (a training or conduct issue).
3. This trend is a required input to the CDS periodic review defined in <<FILL: SOP-ID for periodic review of computerized systems>>, not a standalone activity that stops at the monthly report.

6. Acceptance criteria

A result processed under this procedure is acceptable when all of the following are true:

• The locked processing method was applied first, and its result is known and recorded before any manual integration is considered.
• Every manual integration has a contemporaneous, specific, method-based reason, a completed justification record, and independent second-person acceptance.
• The full reprocessing chain was reviewed in the audit trail, not only the final reported version, with no unexplained migration toward a passing result.
• Every unfavorable locked-method result that lacked a pre-existing scientific basis for override was escalated to OOS or deviation, not reprocessed around.
• Manual integration and reprocessing rates are trended and reviewed at the defined frequency, feeding the CDS periodic review.

7. References

21 CFR 211.68 (automatic, mechanical, and electronic equipment), 211.160(b) (scientifically sound laboratory controls), 211.192 (review of records and investigation of discrepancies), 211.194(a) (complete laboratory records, including all data derived from tests). 21 CFR Part 11 (electronic records and electronic signatures). EU GMP Annex 11 (computerised systems). FDA, “Data Integrity and Compliance With Drug CGMP, Questions and Answers,” guidance for industry, December 2018. PIC/S PI 041-1, “Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments.” USP General Chapter <621>, Chromatography, for system suitability and permitted method adjustments.

Confirm the current version and clause numbers of each reference before issue.

8. Records generated

9. Revision history

10. Approvals

Filled specimen

The following shows how the procedure’s decision points were applied to one illustrative run, so you can see the level of detail an inspector expects behind “we followed the SOP.” The company, system, and numbers are illustrative; replace them with your own.

This is what a defensible, escalation-crossing manual integration looks like under this procedure: locked method result recorded first, a pre-existing documented reason, second-person review with the before-and-after visible, and a trend that shows this pattern is method-driven rather than analyst-driven. Contrast with the same numbers reached by an analyst who reintegrated three times with no documented reason until the number passed. That version has no justification record, no pre-existing scientific basis, and would fail step 5.3 and trigger escalation under section 5.5 instead of acceptance.

Common inspection findings this SOP prevents

• Manual integration performed with no documented reason, or a reason recorded only after the fact once the passing value was known.
• A reprocessing chain where the result migrates toward the specification limit across several versions and only the final, passing version is reported or reviewed.
• An unfavorable locked-method result reprocessed repeatedly instead of being escalated to OOS or deviation.
• No mechanism connecting individual manual integration events to a laboratory-level trend, so a rising rate on one method or one analyst is never surfaced until an inspector finds it first.
• A processing method that is technically “locked” in name but editable by analysts in practice, because the configuration was never verified.

How to adapt this SOP

1. Set your document number, owner, and effective date in the header, and point the deviation, OOS, audit trail review, and periodic review cross-references in sections 2, 5.4, 5.5, and 5.6 to your real procedures.
2. Attach or reference your own list of documented, foreseeable integration situations per method (co-elutions, shoulders, known artifacts) in the method appendix that section 5.2 relies on.
3. Set your own trending frequency and outlier threshold in section 5.6 based on your laboratory’s historical manual integration rate; a fixed numeric threshold works better than “unusually high” once you have a baseline.
4. If your CDS enforces manual integration flagging or method locking natively, cite the specific configuration and its validation record rather than restating the general control.
5. Confirm every regulation in section 7 against the current published version before issue.