FDA 21 CFR 812.28 in LATAM: foreign clinical trial inspectability for device sponsors

21 CFR 812.28 lets FDA use well-designed OUS device data — and inspect the site. Build LATAM FIH for English-speaking inspectability, not ethics stamps alone.

FDA’s foreign clinical data rule is not a slogan. Under 21 CFR 812.28, the agency can use a well-designed, well-conducted investigation outside the United States to support an IDE or a device marketing application — and it can also show up at the site to validate the file. The question US MedTech teams should ask before first patient in Latin America is not “will FDA accept OUS data?” It is “could an English-speaking inspector reconstruct what happened here?”

I am Julio Martinez-Clark, CEO of bioaccess®. This page is the inspectability cut of 812.28 for LATAM device trials. For the IDE-package framing, use Can OUS first-in-human data support an FDA IDE submission?. For country clocks, use the published Panama and El Salvador hubs — not a new timeline invented here.

What 812.28 actually requires

Section 812.28 (final rule, 83 FR 7386, 21 February 2018) says FDA will accept foreign clinical information for an IDE or a device marketing application (PMA, HDE, 510(k), or De Novo) when three conditions are met:

Good clinical practice. Design, conduct, monitoring, auditing, recording, analysis, and reporting that keep data credible and protect subjects — including independent ethics-committee review before initiation, continuing review, and documented freely given informed consent. FDA has stated that conformance with ISO 14155:2020 will generally satisfy the GCP requirement of 812.28 (already on our FDA acceptance guide). Investigations initiated on or after 21 February 2019 must conform to the 2018 foreign-data framework.
Supporting information in 812.28(b). For a significant-risk device under 812.3(m), submit the full (b) set: investigators and sites; protocol and results; a statement that the investigational device is identical to the US device or a detailed comparison; valid scientific evidence under 21 CFR 860.7 if you claim safety and effectiveness; IEC identity meeting 812.3(t); consent, monitoring, and investigator GCP training.
Inspectability. FDA can validate the data through an onsite inspection or other appropriate means if the agency deems it necessary. A LATAM file that cannot be inspected is not an 812.28 file.

Section 812.28(e) is the residual clause: even when a foreign study does not fully meet paragraph (a), FDA may still accept the information if it believes the data are credible and accurate and that subject rights were adequately protected. Do not plan to live in (e). Build (a).

The failure mode I see: ethics approval without an inspectable record

Sponsors who only chase local ethics-committee approval treat the stamp as the finish line. That buys enrollment speed. It does not buy an IDE conversation.

Three patterns burn the file:

Source that cannot be reconstructed. Paper charts in a language and filing system nobody planned to reconcile. When the IDE questions arrive, you cannot show who saw what, when.
Device accountability that dies at customs. Investigational units imported through a commercial distributor “because they already import,” with no chain that survives explant, quarantine, and an English-speaking inspector.
Consent written in English and “translated later.” 812.28 wants documented consent the IEC actually approved. For FDA use, include the 21 CFR 50.25 elements. Spanish first.

Those are the same failure modes already named on the OUS-IDE page. Inspectability is the operational layer underneath them.

What “inspection-ready” means at a LATAM site

Architect the study so an FDA inspector (or a CRO monitor acting for a later US filing) can walk the site without a scavenger hunt:

Electronic data capture with audit trails, not spreadsheet science.
Source documents that map to CRF fields in a single reconciliation plan — imaging, device logs, AE narratives.
Device accountability from import permit through implant/explant that matches the investigator brochure identity claim in 812.28(b)(5).
Monitoring reports that show who visited, what was queried, and what closed — not a one-page “visit done.”
Investigator GCP training on file before first procedure, not after the first query.
eTMF structure that an English-speaking reviewer can navigate: protocol versions, IEC approvals, consents, safety letters, delegation logs.

Panama’s Decreto Ejecutivo No. 21 of 23 April 2026 (Gaceta Oficial No. 30510-C) already puts high-risk protocols through parallel MINSA + Type II ethics review, RESEGIS registration before start, and 24-hour / 15-day serious-adverse-event clocks. That decree structure helps the local file. It does not replace the 812.28 inspectability design. Same rule for El Salvador under DNM/SRS and for any other lead FIH geography we publish: local authorization is necessary; FDA-usable documentation is a separate design choice.

Site selection is an inspectability decision

Pick sites and principal investigators who have already run device protocols with monitors in the room. Early-feasibility n in Panama City is typically 5–20 patients — the right size for a Class III first-in-human cohort (already on the Panama Class III FIH page). Rare disease or a larger n belongs in a multi-country plan.

Public programs already on the FDA-acceptance guide show the pattern: Axoft FINESSE (first cases at The Panama Clinic; FDA Breakthrough Device Designation in 2022); Newrotex (ethics approval in Panama; FDA Pre-Sub for a 510(k) using LATAM data); ReGelTec HYDRAFIL (Colombia early-feasibility, then FDA IDE for a US pivotal). Those names are already public. I am not adding unpublished sponsors or devices under development.

Colombia remains a core market-access geography. Because INVIMA clinical-trial approval timelines have become unpredictable, bioaccess® does not currently recommend Colombia for new first-in-human trial execution. Keep commercial registro and investigational CTA tracks apart.

Pre-Sub before you lock endpoints

If the LATAM protocol is meant to support a later US IDE, file a Q-Sub / Pre-Sub before you lock endpoints (written feedback in 75 calendar days — already on the FDA-acceptance guide). Endpoints and inclusion criteria have to be relevant to the intended US population; site standard of care has to be comparable to US practice. Changing the device after first implant without a comparability memo makes 812.28(b)(5) unforgiving.

One-page gate before first patient

US filing intended. IDE, 510(k), De Novo, PMA, or HDE — named before country pick.
Device identity claim. Identical to the US unit, or a written comparison with change control.
IEC + local authority path. Type II / national desk as the country requires — not an ad-hoc hospital chat.
Spanish consent with 21 CFR 50.25 elements if FDA use is the plan.
EDC + eTMF + device accountability that survive an English-speaking inspector.
Monitoring plan with query closure ownership before first implant.

If you are staring at a LATAM first-in-human calendar and a future FDA conversation, send bioaccess® the protocol stage, device risk class, intended US filing, and whether the investigational unit is identical to the US unit. We will tell you whether the file is being built for 812.28(a) or for a case series.