SystemIT & Data

Data Integrity Assessment & Remediation

A data integrity assessment that draws each GxP process as the path its data really takes, scores every hop on the nine ALCOA+ letters with evidence, reconciles what instruments hold against what was reported, and closes each gap only on signed proof.

See one lab process assessed screen by screen ↓
Watch with transcript →
The problem

The SOP calls the printout the original, and the titrator keeps its own

Data integrity is usually assessed one computerized system at a time, with a questionnaire. A standalone instrument plus a paper worksheet is a hybrid process that rarely makes the system list, and the questionnaire's answer to whether it has an audit trail is that the printout is the original record. The analyst prints the run they are happy with, types the number into LIMS, and a second person checks the printout against LIMS.

Nobody compares what the instrument still holds in memory with what was reported, so a run that never reached LIMS stays invisible until the memory overwrites it. Remediation sits in a tracker with no link to the attribute it repairs, closure is a memo from the owner, and the path is not looked at again until the next cycle or the next inspector.

Built forData integrity leadsQC lab headsPlant QA headsSystem ownersHeads of quality

The Problem Today

A lab process passes its system questionnaire while the instrument still holds results nobody reported.

Assessed from paper today

  • Checklists
  • Printouts
  • Trackers
  • Memos
  • Instrument memory never readOlder results are overwritten before anyone compares them
  • Gaps closed on a memoNobody opens the proof, so the same gap returns

The Solution

  1. The real path, hop by hopDrawn from SOPs, configuration exports and the analyst's walk-through
  2. Nine letters on each hopALCOA+ lit or broken, each letter opening its passage
  3. Hidden copies reconciledInstrument memory laid beside LIMS and the logbooks, row by row
  4. Each gap closed on proofSomeone other than the owner checks the proof; the path is rescored

Built-in controls

  • An unexplained result goes to the lab head and QA
  • The owner of an item never signs its closure
  • People sign the evidence; no agent gives a verdict

What You Get

Data integrity becomes a standing program with proof instead of a push before each inspection.

  • Found by the plant, not an inspectorHidden copies reconciled for each instrument that keeps results
  • Client auditors see the proofEach gap with its interim control, its fix and the signed closure
  • One ruling, applied alikeA QA ruling on one instrument class becomes a rule once signed
Features

Data integrity assessment of one water determination in a QC lab

The data integrity lead, the QC lab head and the plant QA head each work their part, from the first run of the agents to a signed plan with its interim controls.

demoPlants, clients, people, instruments, batches and results shown here are sample data.

  1. 01Data path

    Data flow map of a hybrid lab process, every hop scored on ALCOA+ letters

    • The path as the data really runs. Sample login, titrator, printout, hand transcription, second-person review and certificate show where the record actually goes, paper hops included.
    • Nine letters per hop, lit or broken. Attributable to available, each scored with the passage behind it, so the lead reads evidence instead of filling a questionnaire.
    • The hop where the record leaves the instrument. The step that matters most is marked on the map, with the reconciliation counts directly beneath it.
  2. 02Letter evidence

    Each broken ALCOA+ letter opens the SOP passage behind the score

    • One click from letter to passage. The QA head checks why Original is broken against the SOP's own words before signing anything.
    • Opening is recorded per person. Each signer's opened records are their own, so a signature shows that this person read the evidence.
  3. 03Inspector

    Inspector's questions on the printout, the explained rows and a scoring slip

    • The original-record question answered with the rule. An instrument that stores results makes the electronic result the original, so the lab hears the inspector's argument first.
    • The agents' own scoring challenged. The inspector agent rechecks the scorer's letters against the hops upstream, so a letter lit on a weaker source is corrected before anyone signs.
    • No point skipped. The assessment cannot be signed while an answer waits for a person to accept it.
  4. 04Hidden copies

    Instrument memory reconciled against LIMS, logbooks and lab incident records

    • Memory beside what was reported, per instrument. The QC lab head sees which titrator holds results that never reached LIMS, and how far back its memory still reaches.
    • Explained only by a matching record. A row counts as explained when a logbook, reagent log or incident record shows the same instrument, minute and result.
    • Every unreported row listed with its sample. Product, client and result sit on each row, the scope an investigation starts from.
  5. 05Memory export

    Three runs on one batch in the titrator's memory, one of them reported

    • The memory export as the instrument wrote it. Date, time, sample, result, weight and volume come straight from the exported memory, highlighted where they matter.
    • Printout, worksheet, LIMS entry and memory side by side. The lab head reads all four artifacts the agents read before deciding what happens next.
  6. 06Raise event

    Possible data integrity event raised by the QC lab head, no agent's account

    • Raised by the process owner, inside the quality system. Unexplained results are investigated inside the quality system, as FDA's data integrity questions and answers ask, and QA is notified at once.
    • No explanation offered by the agents. Why a run went unreported is left to the investigation, so nobody's account is written for them.
  7. 07Clients told

    Each affected client told its own unexplained rows under its quality agreement

    • Rows grouped by the client they belong to. Each client receives only its own rows, and rows tied to no single client stay inside the investigation.
    • QA signs each client's notice. The plant's QA head opens that client's rows before signing, so a client hears it from the CDMO first.
  8. 08Remediation plan

    Data integrity remediation plan with interim controls, owners and proof

    • Interim controls that start the next day. A daily memory check and archive export run while the connection to LIMS waits for change control.
    • Each item names the letters it repairs. A fix links to the hop and attribute it closes, and those letters stay broken until its proof is signed.
    • The supplier asked in fixed fields. A structured request to the instrument supplier can change the plan only through its returned fields.
  9. 09Closure

    Remediation items closed only once someone besides the owner checks the proof

    • Proof matched to what the item asked for. The closure checker compares the submitted export with the item's own requirement and quotes where it was found.
    • The path re-scored the same day. Once a second person signs, the letters relight on the data path, so the program shows the repair as well as the closure.

The lead maps, the lab head owns the process, QA signs plans and closures

One person runs the program across plants; each lab owns its paths; the quality unit decides what is fixed.

Data integrity leads

Run the assessment program across plants and keep the evidence pack ready

  • Paths drawn from SOPs, exports and the bench walk-through
  • The riskiest process assessed first
  • A program view for any auditor
QC lab heads

Own their lab's data paths, raise lab events and own the fixes

  • Unreported runs found before an inspector
  • The four artifacts side by side
  • An event the agents never explain
Plant QA heads

Sign assessments, remediation plans, closures and the plant's rules

  • Proof opened before any item closes
  • Client notices limited to each client
  • Past rulings shown when a case recurs
System owners

Own LIMS and instrument interfaces and supply the proof of a fix

  • Fixes tied to the letters they repair
  • Supplier replies limited to set fields
  • Configuration exports accepted as proof

A reconciler counts the unreported runs and is barred from explaining them

Seven agents read records and propose; the people named on each screen decide.

Path drawer

Draws each hop from SOPs, the system inventory, configuration exports and the walk-through note.

  • Draws no hop that no record supports
ALCOA+ scorer

Lights or breaks the nine letters on each hop, each with its passage.

  • Declares no process fit for use
  • Scores nothing without a cited record
Reconciler

Lays instrument memory and local copies beside LIMS and the logbooks, and counts what a record explains.

  • Explains no row without a record
  • Judges no reason
Inspector

Plays the inspector against the SOP, the path and the agents' own work.

  • Decides nothing
  • Leaves no question without an answer or a person
Remediation planner

Proposes interim controls and fixes with the letters each repairs, an owner, a date and the proof that closes it.

  • Assigns no item without its owner named
  • Closes nothing
Outreach agent

Sends structured requests to instrument suppliers and reads the returned fields.

  • Lets no free text change an option
Closure checker

Checks that submitted proof matches what the item asked for, and re-scores the path after a person signs.

  • Signs no closure
More screens

A plan for one lab adds five items to a queue of nearly four hundred

Beyond one process: the open items across plants, the inventory ranked by data risk, the program, the rules and the trail.

Overdue items first, across five plants397 open remediation items, overdue first, each showing its owner, due date and clock, filtered by plant; the QA head can escalate selected overdue items with a signature.
Processes ranked by data criticality and vulnerabilitySystems, hybrid processes and simple instruments ranked by data risk, with hand-transcription steps flagged and the next assessment one click away.
One program view for the head of qualitySystems, paper or hybrid processes, assessed this cycle, open and overdue items for each plant, sent to the head of quality each month.
An earlier ruling on balances, proposed as a ruleA QA head's ruling on balance printouts becomes a proposed Plant 5 rule on when a printout is the original; a rule for every plant needs the head of quality.
Who opened what before they signedEach record a signer opened, the event raised and the plan signed, with the meaning of each signature, downloadable as a CSV file.

demoSample data, from the working solution.

No agent explains a missing result, signs a plan or closes a gap

Explaining an unreported resultAn unexplained row goes to the QC lab head and QA as a possible data integrity event; the investigation decides why.
Judging a batchWhether a result or a batch is acceptable stays with the QC lab head and the plant QA head, and release to market with the client.
Calling anything compliantThe output is evidence and drafts that named people sign.
Closing an itemSomeone other than the owner reads the proof and signs.
Applying a ruleRules learned from rulings take effect only once the plant QA head signs, and one for every plant also needs the head of quality.

SOPs, instrument memory exports and the analyst's own walk-through

SOPs and the system inventoryThe written procedure for each step, and the list of GxP systems and processes.
Instrument memory and configuration exportsStored results, users, firmware and clock settings, exported by the instrument engineer.
The analyst's walk-through noteHow the work is really done at the bench, step by step.
LIMS results and their audit trailWhat was reported, by whom and when.
Worksheets, printouts and logbooksPaper records, reagent logs and lab incident records that can explain a run.
Certificates of analysisThe last hop, where a client's own filing and release rely on the result.

What QA and lab teams ask about data integrity assessment

What is a data integrity assessment?

A data integrity assessment checks whether a GxP process's records stay attributable, legible, contemporaneous, original and accurate, and complete, consistent, enduring and available, from where the data is created to where it is used. Here agents draw the real data path, score each hop with evidence and reconcile what instruments hold against what was reported.

How is a data integrity risk assessment prioritized?

By data criticality and vulnerability, as PIC/S guidance describes. The inventory ranks each system, hybrid process and simple instrument on both, so the riskiest process, such as a standalone instrument with hand transcription, is assessed first.

Is a printout ever the original record?

Only for very simple instruments that do not store data, in PIC/S terms, and FDA accepts a printout only when it is the original or a true copy. An instrument that keeps its own results has an electronic original, which is why the printout hop stays broken until that result is kept and reviewed.

What happens to a result found in memory but not in LIMS?

If no logbook, reagent log or incident record explains it, the QC lab head raises a possible data integrity event inside the quality system and QA is notified. The agents never suggest a reason, and each affected client is told its own rows under its quality agreement.

How does data integrity remediation close?

Each item names the letters it repairs, an owner, a date and the proof that closes it. The owner submits the proof, a second person opens it and signs, and the path is scored again the same day.

Does it cover hybrid systems and paper?

Yes. A standalone instrument with a paper worksheet is assessed by the path its data takes, hop by hop, which is where a system-by-system questionnaire usually stops.

Bring one instrument
that prints its results.

Send its SOP, a memory export and the matching LIMS results; we draw its path and reconcile what it holds.