OOS and OOT investigations that start with the evidence already checked and end in a cited report
Every out-of-limit lab result starts with its evidence already checked and ends in a cited report the supervisor signs.
See one case, screen by screen ↓When a result falls out of specification, the lot goes on hold and two clocks start: five working days for Phase I in the lab, thirty calendar days to close. Before anyone can form a hypothesis, the QC supervisor has to pull the run from the chromatography data system, read the audit trail for the run window, check the bracketing standards, the instrument’s calibration and maintenance, the reference standard and reagents, the analyst’s training on the current method version, interview the analyst and recalculate the result.
Then the rules of the FDA OOS guidance apply: no averaging the original result with retests, no outlier test to invalidate a chemical assay, retests fixed before testing, and invalidation only on a documented test event that reasonably caused the result. Under a deadline, that is where the risk sits — lab error accepted without the evidence, or a real result tested away. And a pattern, such as the same pump on the same HPLC failing again, stays out of sight when every investigation is read on its own.
Hours per Phase I by hand and with the agents — the lab’s own estimate, shown on the solution’s dashboard.
Seven specialist agents gather the evidence, fill the checklist, draft the report and watch the trends; the QC supervisor decides and QA approves.
A Corventa 20 mg tablet lot assays at 93.1 % against 95.0–105.0 % on HPLC-07. Here is what happens next, screen by screen, in the working solution.
Lena opens the lab bench for Westbrook and Leiden: 2,914 results reported in the last 30 days, 7 open investigations (4 OOS, 3 OOT), 3 with Phase I due this week and 3 waiting for her decision. Each reportable result is a dot inside its specification; OOS-26-0147 sits below the lower limit.
Agents at work: “Lab-conditions checker read the HPLC-07 log, standards and training for OOS-26-0147.”
OOS-26-0147: assay by HPLC for lot CV26-0412, method TM-CV-021 v6, run CV-ASY-1007, analyst Sam Patel. Release is on hold with the lot due to ship Oct 14. It is day 2 of 30 with Phase I due Oct 13, and the evidence — audit trail, injection sequence, instrument log, training record, interview note, preparation worksheet and more — is gathered and listed.
The raw-data reviewer reads 64 audit trail entries and finds two pump events and a bracketing standard at 97.1 % at injection 14. The lab-conditions checker finds calibration valid but the check-valve service overdue since Sep 30. The similar-event finder counts four pump events on HPLC-07 since June and no earlier assay OOS in 312 results.
“About 40 seconds · read-only on the LIMS and CDS.”
Eleven checks along the supervisor’s steps in the FDA OOS guidance: method and analyst, raw data, calculations, instrument, standards and reagents, method performance, records. Eight pass. Three are findings — the failing bracket, the pump ripple in the audit trail, the overdue maintenance. Lena accepts or changes each one, and any citation opens the exact passage.
A verdict is “finding” only when a cited record supports it.
Transient pump malfunction on HPLC-07 leads at 86 %: ripple alarms at injections 12 and 14, the 97.1 % bracket right after the samples, the overdue service. Sample preparation (8 %), reference standard (4 %) and a true product result (2 %) are shown with the facts that argue against them — weights recalculate exactly, the two preparations agree at 93.0 and 93.2 %, blend uniformity was 99.6 %.
Plan HTP-0147-1: re-inject retained preparations 1 and 2 on HPLC-04 with fresh bracketing standards, two injections each. The criterion is written before any result exists — both preparations agree within 2.0 % and every bracket passes. Lena sends it, and Omar approves it with his e-signature before any testing.
“Re-injection results are investigational: they are never reported and never averaged with the original 93.1 %.”
On HPLC-04 the bracketing standards pass at 99.8, 100.1 and 99.7 % and the preparations give 99.4 % and 99.7 % — the evidence supports an instrument cause. Lena chooses between two conclusions: invalidate for an assignable laboratory cause, or open Phase II. The justification is drafted from the evidence for her to edit, and she signs with her password and the meaning “I propose this conclusion as QC supervisor”.
“A result is invalidated only on a documented test event that reasonably caused it.”
Omar approves with his e-signature, meaning “approved for QA”. The 93.1 % is invalidated and the lot stays on hold: Aisha Bello retests the original sample twice on Oct 8 under QC-SOP-114 §7.2, and those results decide lot CV26-0412.
Six sections in the lab’s template QC-FRM-114-2: reason, Phase I assessment, hypotheses and testing, conclusion, impact on other results and the proposed CAPA. The impact section names the other results at risk: injections 15–17 sat between the failing and the passing bracket, so they are marked for re-injection too. The CAPA services the check-valves and shortens the service interval from 120 to 60 days on all quaternary pumps.
Investigations opened and closed, median days to close, Phase I on time — 12 of 13 inside five working days — and how often a result was invalidated for a lab cause, by cause, lab and test. Lab error should be rare, and every one carries its CAPA.
The same investigation, seen by the four people who carry it — what their week looked like, and what it looks like now.
Seven specialist agents gather the evidence, fill the checklist, draft the report and watch the trends; the QC supervisor decides and QA approves.
Opens an investigation from each OOS or OOT result in the LIMS — product, lot, test, method, specification, analyst, instrument, run — and starts the clocks.
Reads the chromatography run: system suitability, bracketing standards, integration events, processing-method version and the audit trail in the run window.
Checks the instrument (calibration, maintenance, events), the standards and reagents (lot, expiry, standardisation) and the analyst (training on the current method version).
Finds earlier OOS, OOT and lab deviations on the same product, method, instrument or analyst, and the result history.
Fills the Phase I checklist with verdicts and citations, ranks hypotheses with evidence for and against, and drafts the hypothesis-test plan.
Drafts the Phase I or Phase II report section by section with citations, the CAPA proposal and the impact on other results in the run.
Every night fits the stability results to the trend of earlier lots and flags out-of-trend results; finds repeated patterns and pushes them to the team that owns the cause.
Decides; QA approves. The agents propose; a named person decides.
Anyone in the lab can ask about an investigation, a result or a trend — or tell it which rule to change. Every answer is cited.
What happened on OOS-26-0147?
Corventa 20 mg lot CV26-0412 assayed at 93.1 % against 95.0–105.0 % on HPLC-07. The evidence points at the instrument, not the tablets: pump A pressure ripple of 4.2 % at 10:41 and 3.8 % at 10:47, the bracketing standard right after the samples at 97.1 % against the method’s 98.0–102.0 %, and the check-valve service due Sep 30 not done. Calculations, standards and Sam Patel’s training are all fine, and both preparations are kept at 2–8 °C, so a re-injection on HPLC-04 can test the pump hypothesis.
Can we average the retests?
No. The original result is never averaged with retests — it hides variability. When a method itself averages replicate injections into one reportable result, that average is one test, and that is allowed. The setting is locked.
Who can invalidate a result?
Only people. The QC supervisor proposes with an e-signature and QA lab oversight approves — the agents can draft, never decide. Invalidation needs a documented test event that reasonably caused the result; for OOS-26-0147 that is the pump ripple and the failing bracket.
Add a rule: flag bracketing drift over 1.5 %
Done — a new lab rule: flag any bracketing standard recovering outside 98.5–101.5 %. The raw-data reviewer applies it from the next run; the method’s own limit of 98.0–102.0 % still decides validity. It is saved as a new version of the rule set and recorded in the audit trail.
13 screens from the working solution, on its sample data. Pick one to see it large.
Clocks, approvers and lab rules are settings. Two testing rules from the FDA OOS guidance are locked: no averaging retests with the original result, and no outlier tests to invalidate a chemical assay.
| Setting | Default | Choose from |
|---|---|---|
| Phase I due | 5 working days | 3 · 5 · 10 working days |
| Investigation closes within | 30 calendar days | 20 · 30 · 45 calendar days; an extension needs QA approval |
| Who approves hypothesis-test plans and invalidations | Omar Haddad | A named QA approver |
| Retests after an invalidation | 2 retests | 2 · 3 retests, by a second analyst |
| Average retests with the original result | Never | Locked |
| Outlier tests to invalidate a chemical assay | Not allowed | Locked |
| Bracketing standard drift alert | 2.0 % | 2.0 · 1.5 · 1.0 % |
| Out-of-trend detection | Regression control chart, 95 % prediction interval | Regression · by time point · slope control chart; 95 · 99 % |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample lab data · “target” = the design goal, measured in the live solution · “estimated” = the lab’s own estimate in the working solution · Sources: FDA, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production (2022) · 21 CFR 211.192. People, companies and products named on this page are fictional — characters and sample data in the working solution.
Software that takes an out-of-specification or out-of-trend result from the moment the LIMS flags it, through the Phase I laboratory investigation, to a signed conclusion and report. The OOS / OOT Investigation Assistant adds agents that check the raw data and lab conditions, fill the checklist with citations, rank hypotheses and draft the test plan and report.
It reads the CDS run and audit trail, the instrument log, the standards and reagents, the analyst’s training and the result history, then fills a checklist along the supervisor’s steps in the FDA OOS guidance. Each verdict comes with its citation, a finding only when a cited record supports it, and the QC supervisor accepts or changes each one.
No. The agents propose; the QC supervisor signs the conclusion and QA lab oversight approves it. A result is invalidated only on a documented test event that reasonably caused it, and any hypothesis test needs a plan QA approved before testing.
Yes. Averaging retests with the original result and outlier tests to invalidate a chemical assay are locked off, retests are done by a second analyst with the number fixed before testing, and re-injection results under a test plan are investigational and never reported.
Every night the trend watcher fits each stability study to the trend of earlier lots and flags results outside the prediction interval — a regression control chart by default, with by-time-point and slope control charts as options. It also finds repeated patterns across investigations, such as four pump events on one HPLC, and pushes them to the team that owns the cause.
No. The LIMS, chromatography data system, QMS, instrument logbook, learning system and ERP are connected read-only. Every agent step and every human decision is recorded on the investigation’s activity trail.
Yes. When Phase I is due, when the investigation must close, when owners are warned, who approves plans, invalidations and out-of-trend results, the alert limits and the out-of-trend method are all settings. You can also describe a new lab rule in plain words.
The Agentic Solution Engine builds and deploys it from your requirements — your investigation SOP, test methods, report template and a sample of past investigations — and it goes live once every quality gate has passed. We will walk you through it on your own investigations first.
Any batch-record question answered in seconds, cited to the exact page and step.
1-min filmValidation DocumentsValidation protocols and reports drafted in minutes, every value traced to its source page.
1-min filmAsk the PlantBatch analysis that took three months now takes hours, with the query behind every answer.
1-min filmWe’ll run the OOS / OOT Investigation Assistant on a sample of your own past investigations.