Investigational device accountability in a first-in-human study: what FDA and ANVISA expect

The question usually arrives a few weeks before the first shipment: who counts the devices, who signs for them, and what happens to the ones we don’t use? In a first-in-human device study the answer looks like logistics, but it is a regulatory record. When an inspector or an FDA reviewer looks back at an early study, the device accountability log is one of the first places they check, because it shows whether every unit that entered the country can be traced to a patient, a return or a documented disposal.

I am Julio Martinez-Clark, CEO of bioaccess®. This page explains how I set up investigational device accountability for a first-in-human (FIH) or early feasibility device study run in Latin America, using FDA’s 21 CFR Part 812 as the reference most US-bound sponsors already work to and Brazil’s ANVISA RDC 837/2023 as the most explicit Latin American rule. It goes with our pages on import permits for medical devices in Latin America, the ISO 14155 inspectable file and foreign proctors and field engineers in an FIH study. This is operator guidance, not legal or regulatory advice; check the current text of each regulation before you finalize your plan.

One-sentence answer

Device accountability means the sponsor can show, for every investigational unit by lot or serial number, when it shipped, who received it, which participant it was used on or why it was not used, and whether it was returned, repaired or destroyed, with the site keeping a matching record; FDA spells this out in 21 CFR 812.140, and Brazil’s RDC 837/2023 adds that the sponsor may import only the quantity the study needs, may ship only to approved sites, and is responsible for the final disposition of unused product.

What the rules actually require

FDA, 21 CFR Part 812

• Ship only to qualified investigators. Section 812.43(b) says a sponsor shall ship investigational devices only to qualified investigators participating in the investigation.
• Investigator control. Section 812.100 makes the investigator responsible for the control of devices under investigation, and 812.110(c) says the investigator may let the device be used only with subjects under their supervision and may not supply it to anyone not authorized to receive it.
• Site records. Under 812.140(a)(2) the investigator keeps records of receipt, use or disposition: the type and quantity of device, the dates of receipt, the batch number or code mark, the names of everyone who received, used or disposed of each device, and why and how many units were returned to the sponsor, repaired or otherwise disposed of. Section 812.140(a)(3)(iii) adds a record of each subject’s exposure to the device, including the date and time of each use.
• Sponsor records. Under 812.140(b)(2) the sponsor keeps records of shipment (consignee name and address, type and quantity, date, batch number or code mark) and of disposition (batch or code marks of devices returned, repaired or disposed of, and the reasons for and method of disposal).
• End of study. Section 812.110(e) says that at completion or termination, or at the sponsor’s request, the investigator returns any remaining supply to the sponsor or disposes of it as the sponsor directs.
• Labeling. Section 812.5(a) requires the device or its immediate package to carry the statement “CAUTION—Investigational device. Limited by Federal (or United States) law to investigational use,” along with the manufacturer, packer or distributor and the quantity where appropriate.
• Retention. Section 812.140(d) requires these records to be kept during the investigation and for 2 years after the later of the study’s termination or completion, or the date they are no longer needed to support a marketing submission.

Brazil, ANVISA RDC 837/2023

• Import only what the study needs. Article 37 makes the sponsor responsible for importing only the quantity necessary to run the clinical investigation, and Article 74 prohibits bringing in investigational products not listed in the published DICD approval or in quantities beyond the study’s needs.
• Ship only to approved sites. Article 38 limits distribution to the institutions listed in the DICD form and approved by the ethics committees, and its sole paragraph makes the sponsor responsible for the final disposition of investigational products not used in the study.
• Labeling and blinding. Article 36 requires the investigational device, and any placebo or sham, to be made under good manufacturing practices and coded and labeled so as to protect blinding where applicable and identify it as an investigational product.
• Import file and outer packaging. Articles 70 to 72 route the import through a SISCOMEX sanitary-clearance petition that cites the DICD approval number, and require the outer or transport packaging to show that number, special storage conditions such as temperature, humidity and light, and the lot, identification code or serial number that allows traceability.
• No commercial use. Article 75 prohibits selling these products or changing the purpose of the import.
• Retention. Article 33 requires the sponsor to keep trial data for 5 years after the last approval of a registration or notification of the device in Brazil, or for at least 2 years after discontinuation or completion if no registration follows.

Other Latin American regulators have their own import and reporting forms. In practice, a log built to satisfy 812.140 and RDC 837 at the same time is a sound baseline for Colombia, Panama, Mexico or Argentina, with local forms and language requirements added on top.

Why this is harder in a first-in-human study

• Units are expensive and few. Early builds are often hand-made, serialized and sized in several variants. Losing track of one unit is not a rounding error; it can mean a missing size on surgery day.
• Not every unit is opened for a patient. Some get opened in the room and not implanted, some are used for bench demos or site training, some are damaged in transit. Each of those needs its own disposition code, or the count will never reconcile.
• Explants and returned devices carry data. In an FIH study, a device removed from a participant or returned after a malfunction is often the most valuable engineering evidence you will get. It also has to be handled under the site’s biohazard rules and tracked like any other unit.
• Revisions happen mid-study. If the design or specification changes, Brazil requires the change to be notified to ANVISA (RDC 837, Art. 73 §1), and your log has to keep the old and new versions apart.
• Import quantity is fixed in advance. Because Brazil ties the import to the quantities in the approval, the supply plan has to account for spares, sizes and resupply before you file, not after the first case.

How I set it up

1. One unit, one line. The accountability log has a line per serial number (or per lot for disposables), not per box. Columns: serial or lot, model and size, version, date shipped, consignee, date received, received by, storage location, participant ID or non-use reason, date and time of use, user, disposition code, disposition date, and who signed.
2. Disposition codes agreed before first shipment. Implanted or used; opened, not used; damaged in transit; used for training or demonstration (with prior sponsor written approval); explanted; returned for analysis; expired; destroyed at site under sponsor instruction. Each code says what evidence goes in the file.
3. Training units are separate stock. Units for wet labs or proctor demos are labeled and logged as non-clinical so they can never be confused with clinical supply.
4. Chain of custody across the border. The import file, the airway bill, the temperature or shock data if the device needs it, and the site’s receipt are filed together, so the count at customs matches the count at the site.
5. Reconciliation at every monitoring visit. The monitor counts physical units against the log and the case report forms. A discrepancy is a documented deviation, not a correction in pen.
6. A written return and destruction procedure. Before the first case, the sponsor decides which units come back and which are destroyed locally, who pays for the shipment back, and what certificate of destruction the site must provide. In Brazil this is the sponsor’s duty under Article 38.
7. Close-out count. At site close-out, every unit is accounted for, the last returns are shipped or destroyed, and the final reconciliation is signed by the site and the sponsor and filed in both the trial master file and the investigator site file.

Common mistakes

• Shipping to the hospital’s general receiving dock instead of to the investigator named in the approval, and then losing a day finding the box.
• Hand-carrying units in a proctor’s luggage. It breaks the import record and, in Brazil, conflicts with the requirement that imports run through the sanitary-clearance process tied to the DICD.
• Logging “used” for a unit that was opened and discarded, which makes the device-exposure record wrong.
• Letting the site throw away an explanted device because nobody wrote down that the sponsor wanted it back.
• Importing extra units “just in case” beyond the approved quantity, and then having no lawful way to use or sell them.

How bioaccess® runs this

In the first-in-human and early feasibility studies we run in Latin America, the accountability log, the disposition codes and the return or destruction procedure are part of study start-up, built from the same import file the regulator approves. Our monitors reconcile the count at every visit, so at close-out the sponsor has one clean trail from the factory to each participant that it can hand to ANVISA, INVIMA, COFEPRIS, Panama’s MINSA or FDA. When a device later moves to registration, the same discipline carries into the holder role described on our market access page.

FAQ

Does the site or the sponsor own the accountability log?

Both keep records. Under 21 CFR 812.140 the investigator keeps receipt, use and disposition records and the sponsor keeps shipment and disposition records; the two should match unit for unit.

Can we destroy unused devices at the site?

Yes, if the sponsor directs it. Section 812.110(e) lets the investigator return remaining supply or dispose of it as the sponsor directs, and the sponsor must record the reason and method of disposal. Get a written destruction record from the site.

How long do we keep the records?

For FDA, 2 years after the later of study completion or the date the records stop being needed for a marketing submission. For Brazil, 5 years after the last registration approval in Brazil, or at least 2 years after the study ends if no registration follows.

Can we import a few extra units for training?

Plan it before you file. In Brazil the approved import quantities define what may enter the country, so training units need to be in the request rather than added later.

What label does an investigational device need?

For FDA, the 812.5 caution statement and the manufacturer’s details. In Brazil, RDC 837 requires the device to be identified as an investigational product and the transport packaging to carry the DICD number, storage conditions and a lot or serial number. Local language requirements can apply too, so check them country by country.

Sources: 21 CFR 812.5, 812.43(b), 812.100, 812.110(c) and (e), 812.140(a), (b) and (d); ANVISA RDC 837/2023, Arts. 33, 36, 37, 38, 70 to 75.