WorkflowManufacturing & Sterilization

Device History Record Review by Exception

Device history record review that reads each page against the spec in force on the build date, and stops a serial only where its record fails.

See the demo Book a demo

Follow 24 tables from 08:02 to release at 08:26 ↓

Watch with transcript →

The problem

Quality reads a whole history record to find the one line that is wrong

Finished tables sit in quarantine while a reviewer reads each serial’s record against the device file and each signature against the procedure. Most of those pages are clean, so most of that reading finds nothing, and the release clock runs while it happens.

The misses hide where a tired reader rarely looks. A value passes today’s limit but not the limit in force on its build date, because the traveler was printed on the old form. A good reading came from a tester whose calibration lapsed the week before. One verifier box stays blank among dozens of signatures.

Built forQA release specialistsRecord reviewersProduction supervisorsSite quality directors

The Problem Today

Every finished device waits while quality reads its whole history record, though most pages hold nothing wrong.

400K: lot and serial records closed yearly by one large maker our estimate

Checked by hand today

Travelers
Test prints
Revisions
Calibration

Old limits on old formsA form printed before a change still says pass
Lapsed testers read fineA good number from an expired tester looks good

The Solution

Read each pagePaper travelers, test printouts and label prints, value by value
Use the spec of the dayEach value against the revision in force on that serial’s build date
Check the tester’s dateEach test instrument’s calibration on the day of the test
Stop only what failsA stopped serial goes to the person who can fix it, with its page

Built-in controls

Agents never release, hold or close a serial
The reviewer opens a sample before signing
Release takes a second person and an e-signature

What You Get

Clean devices leave quarantine the morning their records arrive, and a named person decides each stop.

$4.8M: yearly at one large maker, in hours, failures and cash our estimate

A release file people signedOne row per serial, with its signer and its reason
Evidence an auditor can followWho opened, decided and signed what, and why
Each tester checked on the dayA lapsed calibration found before release, not after

One work order, one morning

Device history record review on one work order: 1,136 pages, three stops

Twenty-four powered surgical tables, built over a week and packed, wait in quarantine for their records. The last scanned page lands at 07:38, and this is what happens next.

demoThe plant, its tables, the people and each reading are sample data.

0108:02
1,136 pages in for 24 tables, and not one of them read yet
QA release specialist
The last page reached the scan station at 07:38: 848 traveler pages plus test printouts, exports and label prints. The route card lays out the 19 stations of the routing with each serial still in the pages-in bin. Release is due within two working days of the last operation, and the clock shows 34 hours left.
0208:06
Twenty-one serials reach ready for QA, and three stop where their records fail
QA release specialist
Seven agents read the pages and walk each serial’s tag along the routing. Twenty-one reach ready for QA. Three stop with a reason chip: 0422 at torque verification with a signature missing, 0409 at the lift load test, already asked of the production supervisor, and 0417 at electrical safety, tester out of calibration. Two need her decision.
0308:06
Two tables drifted 4 mm, and only one of them is out of limit
QA release specialist
The spec in force view puts each serial on its kitting date. Rev D of the device file took effect the day 0409 was kitted and tightened lift drift to 3 mm or less. Its traveler was still the rev C form, which printed 5 mm and said pass. Serial 0403 also drifted 4 mm, built under rev C, and is correctly clean.
0408:06
Each value within limit, and still a stop: the analyzer was past its due date
QA release specialist
Earth leakage 212 µA against a 500 µA site limit, touch current 38 µA: the printout passes. The safety analyzer was due for calibration two days before this test ran, and the other 23 serials were tested on a second analyzer. The agents’ plan ends in a proposal: hold for a retest on that one, which the test technician can run today.
0508:06
The calibration record: out at the lab, with its as-found result not back
QA release specialist
One citation opens the calibration record as the calibration system exported it: a 12-month interval, due two days before the test, out for calibration since the day after it, as-found result not back. Below it the panel notes that earlier results must be assessed when equipment is found not to conform. Close with a reason stays shut until that result is in.
0608:13
The reviewer opens three sampled clean serials, then signs for all 21
Device history record reviewer
The sample is drawn at random when the read finishes: 10% of the clean serials, at least 3, here 0403, 0412 and 0420. His sign button stays shut until each one is open. He signs the review of the 21 with his password at 08:13, and the release step lists them ready, noting that he cannot release what he reviewed.
0708:21
A blank verifier box cannot be filled after the fact, so 0422 goes back
QA release specialist
Four column-to-base bolts were torqued within spec, but the verifier box on traveler page 22 is blank. She has already held 0417 for a retest, which takes effect at once. For 0422 she picks the reason code, second-person verification missing, and signs it back to production for a witnessed re-check and a late entry.
0808:26
Twenty-one tables released the morning their pages arrived
QA release specialist
She opens one clean record herself, as the settings require, then releases the 21 under her e-signature, meaning released for distribution. The route card moves them to the released bin. Three are still out, each with its owner: 0409 with production, 0417 held for a retest, 0422 sent back. The agents release nothing and hold nothing.
0908:26
Only a release signature moves a serial out of quarantine in the ERP file
QA release specialist
The release file the ERP loads holds one row per serial: release or keep in quarantine, the reason, who reviewed, who released and the signature meaning. The router rebuilds it after each decision. Twenty-one rows read release, and 0409, 0417 and 0422 stay in quarantine with their reasons, so no serial leaves on a default.

Same 4 mm of drift, two build dates, two different answers

Six tables were kitted under rev C and eighteen under rev D. Each value is held to the limit in force on its own build date.

Serials	Revision on the build date	Drift against the limit	Result
0403	Rev C	4 mm against 5 mm or less	In limit, not stopped
0409	Rev D	4 mm against 3 mm or less	Over the limit, stopped
0401 to 0406	Rev C	Six serials on the old limit	Held to rev C
0407 to 0424	Rev D	Eighteen serials on the new limit	Held to rev D

demoSample data, from the working solution. Against today’s rev D, 0403 would fail. It was built before rev D, so it is not flagged.

About 18 reviewer hours a work order, down to about one

Our estimate for this work order, from line-by-line reading against the records a person opens here.

Serials in the work order	24
Reading one record line by line	45 min
Reading the whole work order by hand	about 18 h
Records a person opens here3 sampled clean serials, 2 stops, 1 more clean record	6
At about 10 minutes each	about 1 h

estimatedOur estimate.

Around the release

The rest of the record, the floor’s questions and the release trail

What the reviewer, production and the site quality director see on the same work order.

One serial’s device history record by station: values with limits in force, instruments in calibration, signers and source pages — **One serial’s record**Station by station: each value against the limit in force, the instrument and its calibration on the test date, who signed and who verified, and the source page.

Label and UDI check for a work order: 24 labels matched on device identifier, manufacture date and serial against the record — **Label and UDI check**Each printed label against the record and the label spec: device identifier for the configuration, serial and manufacture date. All 24 match here.

Production and test questions: a lift drift question, a retest on the in-calibration analyzer and a witnessed re-check — **Production and test**One question per serial, each with its page, for production and test. Each answer is checked again by the agents before anyone can release.

Who agreed panel after release: read by 7 agents, reviewed by the record reviewer, released by QA for 21 serials, 3 stops open — **Who agreed**Read by the agents at 08:06, reviewed at 08:13, released at 08:26, with each open stop and its state. Reviewer and releaser are different people.

Audit trail for the work order: release, send-back, hold, sampled openings, the signed review and each agent stop with its reason — **Audit trail**Each read, opening, decision and signature in order, people and agents alike, with the reason and the signature meaning. The site quality director can export it.

Settings for stop rules, sampling and release clock, approvers with separation of duties, and systems of record connected read only — **Stop rules**What stops a serial, the sample and the release clock. Calibration, signature and unreadable-page rules are locked on, and a looser change starts on the next work order.

demoSample data, from the working solution.

QA release decides each stop, and the reviewer never releases

Holding or sending back takes effect at once. Closing a stop with a reason waits for a second person.

QA release specialist

Sees: Each stop with its page, its source and the agents’ plan
Signs: Holds, sends back or closes each stop, and releases clean serials

Device history record reviewer

Sees: The sampled clean serials and each typed note in its exact words
Signs: The review of the clean set, and a countersignature on any close

Production supervisor

Sees: One question per serial with the traveler page
Signs: Nonconformances, witnessed re-checks and late entries

Test technician

Sees: Which test to rerun and on which instrument
Signs: Retest results, and no release

Calibration coordinator

Sees: Each record measured on an instrument after its due date
Signs: As-found results from the calibration lab, and no record decisions

Site quality director

Sees: The release clock and the full trail
Signs: Nothing: read only, with exports for audits

The agents

Seven agents read and check each page; none of them releases a serial

Each has one check and hands what fails to the router.

Record reader

Reads traveler scans, test printouts, execution entries and label prints, and maps each value and signature to its serial and station.

Never guesses an unreadable value

Spec-in-force checker

Finds the device file revision in force on each build date and the traveler form actually used.

Never applies today’s revision to an earlier build

Value checker

Checks each recorded value against the limit in that revision.

Never rounds a value to pass
Never accepts pass without the number

Instrument checker

Looks up each test instrument’s calibration status on the test date.

Never accepts a value from an instrument out of calibration on the test date

Signature checker

Checks performer and second-person signatures, dates, corrections and late entries.

Never signs, initials or fills a blank

Label and UDI checker

Compares each label print with the record and the label spec.

Never prints or reprints a label

Router

Sends each stop to the person who can fix it with the page, keeps the clock and rebuilds the release file.

Never releases, holds or closes a serial

Seven systems a release reads, and one file it writes

Each system is read in place. The only thing written is the ERP release file, one row per serial.

Manufacturing execution systemTest-station entries and work orders

Scan stationPaper traveler pages as they arrive

Quality management systemNonconformances, as signed references

Product lifecycle managementDevice file revisions and change orders

Calibration management systemInstrument due dates and certificates

Label printing systemThe label print log

ERPReceives the release file; serials leave quarantine only on a release row

What QA release teams ask about device history record review

What is device history record review by exception?

Device history record review by exception means people read only what fails. Agents read every page of each serial’s record, check each value, signature, instrument and label, and stop a serial only where something is missing, out of limit, unsigned or measured on an out-of-calibration instrument. A reviewer samples the clean serials and QA release decides.

What counts as an exception?

A missing value, or one out of limit against the revision in force on the build date. An instrument out of calibration on the test date, or a missing or late signature. A label mismatch, an unreadable page, or a traveler form that differs from the revision in force.

Does DHR review by exception work with paper travelers?

Yes. Scanned traveler pages, test-station printouts and label prints are read alongside electronic entries. Handwriting the reader is unsure of is checked against the matching export, and an unreadable page stops the serial with a request to rescan.

Who releases the serials?

The QA release specialist, under her own e-signature with its meaning. The reviewer and the releaser are always different people, and the agents release nothing.

What happens when a tester was out of calibration?

Each result measured on it after its due date stops, however good it reads. The calibration coordinator owns the impact list and records the lab’s as-found result; until it is back, a stop cannot be closed with a reason.

Can a stop rule be relaxed?

The administrator proposes and QA release approves with an e-signature. A looser rule starts on the next work order, a tighter one at once, and the calibration, signature and unreadable-page rules stay locked on.

What about a unit held by complaint handling?

It shows as one signed reference that stays unopened by the agents, and the serial cannot be released from here.

Manufacturing & Sterilization

Bring one work order’s records
and the spec it was built to.

We’ll read each page and show you which serials stop, and why.

Book a demo Browse solutions

Device History Record Review by Exception

Quality reads a whole history record to find the one line that is wrong

The Problem Today

The Solution

What You Get

Device history record review on one work order: 1,136 pages, three stops

1,136 pages in for 24 tables, and not one of them read yet

Twenty-one serials reach ready for QA, and three stop where their records fail

Two tables drifted 4 mm, and only one of them is out of limit

Each value within limit, and still a stop: the analyzer was past its due date

The calibration record: out at the lab, with its as-found result not back

The reviewer opens three sampled clean serials, then signs for all 21

A blank verifier box cannot be filled after the fact, so 0422 goes back

Twenty-one tables released the morning their pages arrived

Only a release signature moves a serial out of quarantine in the ERP file