Buying and sourcing

The adverse-event record for unapproved research peptides is nearly empty

This page describes the size and shape of a public record, and the reporting machinery that produces it. It does not say any compound is safe, does not say any is dangerous, and gives no advice about use — that requires someone who can examine you.

What the queries return

Querying the openFDA adverse event endpoint by product name in the extract dated 28 April 2026 returns 15 reports for BPC-157 and 10 for IPAMORELIN. A query for TB-500 returns no matching records at all — the API responds NOT_FOUND rather than zero, which is the same practical answer with a different technical meaning.¹²³

The comparison that makes those numbers legible is OZEMPIC in the same extract: 57,104 reports. And a peptide that does sit inside the prescription system, sermorelin, returns 58 — small, but not nothing, and the difference is the route to market rather than the molecule.⁴¹¹

Where reports actually come from

Most of what reaches the database arrives through a legal obligation. Under 21 CFR 314.80, applicants holding an approved drug application must review adverse experiences from any source and forward them to FDA on defined timetables. That single duty is the largest inflow into the system.⁵

Everything else is voluntary, through MedWatch, and it typically involves someone who prescribed, dispensed or treated: a physician, a pharmacist, a hospital. Consumers can and do report directly, but a consumer usually reports a product they can name, from a source they can identify.⁶

• No approved application means no applicant, and no applicant means nobody carries the reporting duty.
• No prescription means no pharmacist and no dispensing record connecting a person to a product.
• Sellers labelling a product for research use rather than human use are not filing safety reports about human use.
• A person who bought something outside the medical system may not tell a clinician what they took, and a clinician cannot report a product they were never told about.⁵⁶

The naming problem compounds it. Approved products fail to match across generic and brand strings even when everyone is trying; research-market products are sold under sequence names, catalogue codes and abbreviations that no query anticipates, and FDA acknowledges duplicate and incomplete records among what does arrive.¹⁰

Why the empty file cannot be read as reassurance

Under-reporting is severe even where the machinery works. Hazell and Shakir's review of 37 studies from 12 countries found a median under-reporting rate of 94% for adverse reactions reaching spontaneous schemes, and 85% for specific serious drug–event pairs. Those figures come from settings with prescribers, pharmacists and manufacturers in the loop.⁸

For products sold outside that loop, the equivalent shortfall has never been measured, and there is no way to measure it from the public data. The honest statement is not that under-reporting is worse; it is that the size of the gap is unknown and cannot be bounded.⁸¹⁰

This is the general form of an old point: absence of evidence is not evidence of absence. Altman and Bland's note makes the statistical version of it, and the reporting version is stricter still, because here the absence is produced by a system that was never pointed at these products.⁹

What the regulatory record does contain

Event counts are not the only kind of official record. FDA maintains a list of bulk drug substances nominated for use in compounding that the agency has identified as presenting significant safety risks, and the reasons it gives are documented in its published evaluations. That is a regulatory determination about a substance, produced by review rather than by counting reports.⁷

It is worth keeping the two apart. A substance can appear in that regulatory record while having almost no adverse-event reports, and the two facts are not in tension: one is an assessment of what is known and unknown about a substance, the other is a tally of paperwork that happened to arrive.⁷¹⁰

BPC-157 is the clearest case on the site: high reader interest, no approved US product, and 15 adverse event reports in the current extract. The number is small because there is almost no mechanism to produce a larger one.

TB-500 returns no matching records at all in the adverse event endpoint. That is an absence of reports, not an absence of events, and the distinction is the whole subject of this page.

Common questions