Drug Accountability: The 483 Finding Nobody Talks About

An FDA investigator sits down at your site with the drug accountability log, the dispensing records, and the subject case histories, then starts adding. Received, minus dispensed, minus returned, minus destroyed. If the arithmetic does not close, you have a finding.

FDA’s Bioresearch Monitoring FY2023 metrics list inadequate accountability and control of the investigational product among the five most common clinical-investigator inspection observations, alongside the ones clinical operations teams worry about openly: failure to follow the investigational plan, and inadequate case history records. FDA ran 681 clinical-investigator inspections that year, classifying 136 as Voluntary Action Indicated and 9 as Official Action Indicated. Drug accountability gets less airtime in vendor demos and audit-readiness decks than informed consent or protocol deviations, and it lands on the Form 483 anyway.

One caveat on that list, since other vendors get it wrong: FDA publishes those observations as an unranked set, with no citation counts attached. Anyone telling you drug accountability is the “#4 most-cited” finding is reading a bullet position as a leaderboard. What FDA supports is narrower and still worth your attention: this is one of the handful of things inspectors write up most.

This post is for sponsors, CROs, and clinical operations leaders. It covers what a drug-accountability finding actually reads like, why manual pill counts and paper logs generate them, and how to build a reconciliation an inspector cannot pull apart. It pairs with our broader guide on what inspectors check in clinical trial audit software.

What FDA counts as a drug-accountability finding

The federal requirement is short and specific. Under 21 CFR 312.62(a), “an investigator is required to maintain adequate records of the disposition of the drug, including dates, quantity, and use by subjects.” The same regulation requires the site to keep those records for 2 years after a marketing application is approved, or 2 years after the investigation is discontinued and FDA is notified.

An investigator flags a drug-accountability deficiency when the records fail one of three tests. The quantities do not reconcile across the drug’s lifecycle. The dispensing entries are missing dates or subject links. Or the log claims a dose was given with nothing in the case history showing the participant took it. Each failure is a documentation gap first and a data-integrity gap second, because an inspector reads an unreconciled count as evidence you cannot account for where the study drug went.

The four numbers that must reconcile

Accountability is arithmetic that has to close for every lot, at every site, on demand. Four running totals drive it: what the site received, what it dispensed to subjects, what subjects returned, and what the site destroyed. ICH E6 4.6.3 sets the same expectation for Good Clinical Practice. The investigator, institution, or a designated pharmacist should keep records of the product’s delivery to the site, inventory at the site, use by each subject, and return or alternative disposition of unused product, with dates, quantities, batch or serial numbers, expiration dates, and the unique code numbers.

The number	What it should equal	How the chain breaks
Received	Shipment records from the sponsor or depot, by lot and quantity	A shipment logged late or against the wrong lot number
Dispensed	The sum of every unit handed to a subject, with date and subject ID	A dispensing event entered from memory days later, or not at all
Returned	Unused units brought back at each visit, counted and logged	A bottle returned to a coordinator who never logged it
Destroyed	Units documented as destroyed under an approved procedure	Waste discarded without a witnessed, dated destruction record

Received has to equal dispensed plus returned plus destroyed plus what is still on the shelf. One untracked bottle or one unlogged return, and the equation fails. The inspector does not have to prove the drug was diverted. The open gap is the finding.

Why manual pill counts and paper logs generate these findings

Paper accountability logs fail in predictable ways. A coordinator counts tablets at the end of a long visit day and transcribes the number by hand. Dispensing entries get backfilled from memory. Returned product waits in a drawer before someone logs it. The same counts live in a binder, a spreadsheet, and the site’s own tracker, and the three drift apart. None of these are exotic errors. They are the routine friction of running a busy site with a manual process, and they accumulate into an unreconcilable log.

The deeper problem is what a count can prove. A pill count shows a bottle is lighter than it was at the last visit. The participant may have swallowed the missing tablets on schedule, dumped them in a sink before the visit, or taken four on Sunday to catch up. We wrote about this failure mode at length: a pill count proves a lighter bottle, not a taken dose. A log that reconciles on quantity can still leave the case-history side of 312.62 wide open, because “use by subjects” means ingestion, and a subtraction never captures ingestion.

Monitors run into this at every interim visit. They can confirm the math on a good day, and they still cannot confirm that the 28 tablets missing from a bottle became 28 doses taken across 28 days. The count and the case history describe the same participant. Only one of them describes what that participant did.

What a real accountability 483 or Warning Letter looks like

FDA clinical-investigator warning letters that cite 312.62 tend to follow a pattern you can recognize before an inspection starts. The drug accountability log shows a subject received a quantity of study drug on a date, and the progress notes for that subject say nothing about a dispensing event. Returns get counted at a visit and never reconciled against what was dispensed.

The purest version is the case where no log survives to reconcile. FDA’s December 2025 warning letter to Maria A. Carballosa, M.D. cites failure to retain records required under 21 CFR 312.62(c) for two years, and the missing records include exactly what this post is about: “records of the disposition of the study drug, including dates, quantity, and use by subjects.” She told FDA the records had been transferred, and FDA’s response is the line to remember: “you did not provide any evidence of such notification or agreement.” When the log is gone, the arithmetic cannot be run at all, and the burden of proving it once closed falls on you.

The common thread is traceability. Every unit of study drug should trace to a dispensing event, and every dispensing event should trace to evidence the subject used the dose. When either link is missing, the observation writes itself.

Closing the gap: reconciliation plus dosing evidence

Good accountability does two things at once. It reconciles the four numbers in real time, and it ties every dispensed unit to verified evidence the participant took it. Most systems do the first and skip the second, which leaves the case-history gap open. Here is how a manual approach compares with one built around continuous reconciliation and dose-level evidence.

Dimension	Paper log and pill count	Automated reconciliation plus dosing evidence
Received vs. dispensed vs. returned vs. destroyed	Reconciled at visits, by hand	Reconciled continuously as events post
Dispensing dates and subject links	Backfilled from memory, error-prone	Timestamped at the point of dispensing
Proof of ingestion	Inferred from a lighter bottle	Captured as a verified per-dose record
Audit trail	Editable binder or spreadsheet	Tamper-evident, hash-chained log
Inspection posture	Reconstruct the story after the fact	Hand the inspector a closed reconciliation

Trialsights participant compliance is built for the second column. Smart pill counts keep the received-dispensed-returned-destroyed arithmetic current instead of waiting for a visit. AI video dose verification asks the participant to record a short clip of the ingestion, which the system verifies and timestamps, so the case-history side of 312.62 holds actual evidence rather than an inference. And a hash-chained audit trail makes every entry tamper-evident, so an inspector can confirm no record was altered after the fact. None of this removes the site’s judgment or the monitor’s review. It removes the manual gaps that turn a good site into a 483.

An inspection-ready accountability checklist

Before your next monitoring visit or inspection, confirm you can answer yes to each of these.

• One source of truth for IP counts, so the binder, the tracker, and the system cannot disagree.
• Timestamped dispensing and return records, captured at the event rather than reconstructed later.
• Dose-level ingestion evidence for each dispensed unit, not a subtraction that assumes the dose was taken.
• Automated reconciliation of received, dispensed, returned, and destroyed, run continuously rather than at visits.
• An immutable, tamper-evident audit log that shows no entry was changed after it was recorded.

If any answer is no, that is the gap an inspector will find first. You can see the whole chain working on a live demo trial. Book a demo and we will walk through smart pill counts, video dose verification, and the reconciliation an inspector cannot pull apart.