Invalid case review, every night: each closed safety case re-read against its sources before a serious case can hide behind a closure code
Every case closed as invalid or non-reportable is re-read overnight, so a serious case cannot hide behind a closure code.
See one case, screen by screen ↓Intake closes thousands of reports a month as invalid, non-case, duplicate or non-serious — at a contact centre, in a patient support programme, in medical information. Each closure rests on one person's reading of a call or an email, sometimes under an SOP clause that should not have been there: “reporter says unrelated”, “no consent for follow-up”, “caller would not give the patient's name”.
QC re-checks a sample of 5–10%. The rest stay closed, and the attachment that arrived the next morning — the email with the patient's initials, the discharge summary, the photo of the pharmacy label — is never read against the closure. When a valid serious case is finally found, its Day 0 has not moved. The 15-day clock ran from the day the report first reached the company, and the report may already be late.
Estimated split for one questioned closure, re-checked by hand and with the solution.
Six specialist agents collect, re-read, check and compare every closed case, then draft the reopened report; a PV quality lead decides and signs.
On Sep 30 a woman calls the medicines line about Corventa: her husband was found unresponsive at home. On Oct 2 the case is closed as invalid — no identifiable patient. Here is the second look, screen by screen, in the working solution.
From 22:00 to 02:41 the agents re-read 1,426 closed cases — 1,186 recent closures and 240 from the Q3 look-back. Seven closures disagree with their sources; six of those are serious, and three are serious and not in the label, so a 15-day Alert is owed. Twelve cases wait for her, closest to their deadline first.
“Each clock started on the day the report first reached us, not the day it was flagged.”
NW-26-118204, Harborline call HL-0930-2271. The caller, Ruth, would not give her husband's full name on the phone, and the triage note reads: “Patient name and DOB declined → no identifiable patient. Closed as invalid per SOP-PV-012 §4.2.” The re-reader marks what the transcript does say: a man of 67 called Robert, on Corventa 10 mg for about three weeks, found unresponsive on Sunday the 28th.
From the transcript: “My husband Robert was taking them — he's 67.”
Ruth's email of Oct 1 — the day before the closure — attaches a photo of the pharmacy label and copies it out: “Patient: R.K., date of birth 1959”, Corventa 10 mg, dispensed Sep 6; the photo shows lot CV2604-118. Initials and year of birth, on file, before the case was closed.
Any one patient descriptor makes a patient identifiable.
Side by side, what was coded at closure and what the sources say: identifiable patient “Missing” against “Male, 67, first name Robert; initials R.K., born 1959”; reporter, event and product all agree. Seriousness: death. Expectedness is checked against the Corventa prescribing information, version 4.0 — death is not a listed reaction, so the case is unexpected.
From the label: “Death, sudden death and cardiac arrest are not listed adverse reactions for Corventa.”
The comparer opens a disagreement: Valid — serious, unexpected, 97% sure, flagged by the nightly run. Day 0 is Sep 30, the day the call came in, so the 15-day Alert (US) and 15-day serious report (EU) are due Oct 15 — 8 days left. Every agent step is shown with what it found, ending at “Waiting for you”.
“Day 0 stays Sep 30 — the day the information first reached us — even though the case was flagged Oct 7.”
Dana has three choices: keep closed, reopen for follow-up, or reopen and report. Before she signs, the screen lists what will happen: reopen in the safety database, keep Day 0 at Sep 30, send it to Dr. Maya Chen for medical review, draft the expedited report from the cited passages, send feedback to Harborline. She signs with her password; the meaning is recorded with it.
Meaning of signature: “Valid case — reopen and report.”
Only now does the drafter act. Task RT-118204 opens in the safety database with the criteria and passages pre-filled; Day 0 stays Sep 30, due Oct 15. Medical review goes to Dr. Maya Chen with seriousness and the label check already cited. The E2B(R3) report has 38 fields filled from the sources, and Harborline gets feedback that its “No identifiable patient” closure was overturned.
The drafter drafts — a person submits.
Every second-look decision sits in one log: the case, how it was closed, the decision, the meaning of the signature, the signer, the signature ID and the follow-through. Kept-closed decisions carry their reason — “‘Dr.’ in the email signature is a forum nickname, not a physician”. Late reports carry their deviation number, drafted the day they were found. One click exports the inspection pack.
Over 30 days: 24,880 closures re-read, 217 questioned — 0.87% of closures — and a median of 4.6 hours from flag to signed decision against a target of under 8. Overturned closures are broken down by closure reason, so the reasons that fail most stand out.
“Relatedness and consent closures are overturned almost every time — both clauses were removed from the SOP on Oct 5.”
Each intake partner shows its closures, how many were questioned and overturned, and the rate per 1,000 closures, with an owner and the latest action. Harborline Contact Centre: 71 overturned from 9,120 closures, retraining on patient identifiers booked Oct 9. Wrenfield Patient Support: work instruction WI-PSP-07 withdrawn Oct 5.
The same closed case, seen by the four people who carry it — what their week looked like, and what it looks like now.
Six specialist agents collect, re-read, check and compare every closed case, then draft the reopened report; a PV quality lead decides and signs.
Every night, pulls each case closed as invalid, non-case, non-reportable, duplicate or closed — plus the look-back batch — with all its attachments.
Reads every source — call transcripts, emails, intake forms, discharge summaries, journal articles, photos — and pulls out who, what, which product and when, with the passage.
Applies the four minimum criteria from the validity SOP, assesses seriousness, and checks expectedness against the product label.
Compares the coded closure with what the sources show, works out Day 0 and the regulatory clock, and opens a disagreement for a person.
After a person signs, reopens the case task, keeps Day 0, routes medical review, drafts the E2B(R3) report fields, and drafts a late-report deviation when the clock has run out.
Trends overturned closures by intake partner and closure reason for the monthly quality review and vendor oversight.
Decides; a physician reviews. The agents propose; a named person decides.
Ask about any closed case, a clock, a partner or a rule — or say what to change. Answers cite the source passage.
Why was NW-26-118204 closed, and what does the source say?
It was closed on Oct 2 as invalid — no identifiable patient, because the caller would not give her husband's full name or date of birth. But the sources do identify him: she says “my husband Robert … he's 67”, and her email of Oct 1 gives initials R.K. and year of birth 1959. Any one descriptor is enough. The event is a death, which is not in the Corventa label — so a 15-day Alert report is owed from Day 0 = Sep 30, due Oct 15.
Which disagreements are closest to their deadline?
NW-26-117690 (reporter says unrelated) is 4 days late; NW-26-116955 (no consent for follow-up) has 1 day left; NW-26-117402 (no adverse event) has 2 days left. NW-26-117690 is past due — reopening keeps Day 0 Sep 18 and drafts the late-report deviation.
Add a rule: a lot number in a photo identifies the product
Done. From tonight the Validity Criteria Checker looks up any lot number in a transcript or photo in the release records — as in NW-26-118102, where lot VB26-0844 identifies a Veltrimab pen although the caller never said the name. The rule is on in Settings and recorded in the audit trail.
Which intake partner closes the most valid cases?
Harborline Contact Centre has the most overturned closures (71 in 30 days), mainly missing patient identifiers (26) and “reporter says unrelated” (12). Wrenfield has the highest rate — its 12 consent closures stopped when work instruction WI-PSP-07 was withdrawn on Oct 5.
13 screens from the working solution, on its sample data. Pick one to see it large.
What is re-read, which validity rules apply and who signs what are settings — and the rules your SOP fixes can be locked.
| Setting | Default | Choose from |
|---|---|---|
| Run time | 22:00 every night | 22:00 every night · 02:00 every night · Every 4 hours |
| Closure reasons in scope | 8 of 8 reasons | No identifiable patient · No identifiable reporter · No adverse event · No suspect product · Duplicate · Reporter says unrelated · No consent for follow-up · Non-serious — no expedited report |
| Re-read window | 7 days after closure | 7 · 14 · 30 days after closure |
| Q3 look-back | 240 cases a night | 240 · 500 cases a night · Paused |
| The reporter's view on relatedness never makes a report invalid | Locked | SOP-PV-012 v8 §4.3 |
| Signs the reopen decision for serious cases on a 15-day clock | Dana Okafor | Dana Okafor · Sam Patel · Dr. Maya Chen |
| Medical review after reopening | Dr. Maya Chen | Dr. Maya Chen · Safety physician on call |
| Feedback to intake partners | Lena Ortiz · weekly digest | Weekly digest · each case |
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 safety data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · Context: GVP Module VI, 21 CFR 314.80, and the FDA warning letter on unreported serious cases (March 2026). People, companies and products named on this page are fictional — characters and sample data in the working solution.
A second check of reports closed as invalid, non-case or non-reportable, to make sure none of them actually meets the four minimum criteria — an identifiable patient, an identifiable reporter, a suspected adverse event and a suspect product. Invalid & Closed Case Second Look runs it on every closed case, every night, against every source document, instead of on a QC sample.
The Source Re-reader pulls patient descriptors, reporter, events and products from every attachment with the exact passage. The Validity Criteria Checker applies your validity SOP — any one patient descriptor is enough; relatedness and consent never decide validity — then assesses seriousness and checks the label. Where that differs from the closure, the Decision Comparer opens a disagreement with a confidence and the reason.
No. The agents are read-only on the safety database and never edit a case. They flag; a named person decides. Only after a signed decision does the Reopen & Report Drafter create the reopen task and draft the report, and a person submits it.
Day 0 does not move: it stays the day the information first reached anyone acting for the company. If the clock has already run out, the drafter prepares a late-report deviation with the full timeline for the PV compliance manager, and the QPPV is told.
Yes. A look-back re-reads past closures alongside the nightly run. In the working solution it re-reads all 18,400 Q3 closures against the corrected SOP at 240 a night, with options to speed up to 500 a night or pause.
Every decision is an electronic signature with its meaning, signer and time, in a decision log with the follow-through — report sent, deviation drafted, or the written reason a case stays closed. The log downloads as CSV or exports as an inspection pack.
Yes. Agents only flag and draft. A PV quality lead decides each disagreement and signs, a safety physician reviews every reopened serious case, and rules your SOP fixes — such as relatedness never making a report invalid — are locked.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your validity SOP, product labels, closure codes and a sample of past closures — and it goes live once every quality gate has passed. We will show it to you on your own closed cases first.
Every side-effect report becomes a cited case in minutes, filed with days to spare.
1-min filmSafety LiteratureEvery journal article screened with a written reason, and reportable cases at intake in about a day.
1-min filmSignal ManagementEvery potential safety signal gets its evidence assembled in days, each number traced to its query.
1-min filmWe’ll run a second look over a sample of your own invalid and non-case closures.