624 trials terminated or suspended over safety concerns — and 104 without publicly identified results

How we counted

01From 36,143 stopped trials to 624 confirmed

Every medical student learns that stopping a trial for harm is protection, not failure. What nobody teaches is what comes next: the reason stays in a free-text field of the registry, and the results never reach the literature. We opened ClinicalTrials.gov — the world’s largest public trial registry — and ran this exercise on every stopped study listed there, terminated and suspended:

1. 34,707 trials marked “Terminated” and 1,436 marked “Suspended” in the US registry (as of 11 November 2026). Of these, 27,147 terminated and 1,311 suspended state a reason.
2. We automatically selected the 1,826 terminated + 112 suspended in which the reason touches harm to humans: toxicity, adverse events, deaths, DSMB/IDMC. We explicitly excluded negations (“not due to safety”), COVID-19, and efficacy reasons (“futility”).
3. We read — line by line, manually — all 1,938 reasons. For terminated trials, 619 survived and a systematic rescan removed a further 27 explicit negations and administrative closures. For suspended ones, reading all 112 left 32. 624 in total.
4. For each of the 624 we searched the literature two ways: by NCT ID on EuropePMC and, for those with no posted results and no NCT mention, a secondary search of drug + condition + sponsor. We manually read the titles of every plausible secondary hit: 211 posted results and appear in the literature, 257 appear in the literature but never posted results, 52 are registry-only, and 104 are completely silent.

The register

02The 624 trials — search a sponsor, a disease or an NCT

Romania

0416 trials recruited in Romania — 15 posted or mentioned results, one suspended and silent

Of the 624, fifteen terminated trials and one suspended trial had sites in Romania — 13,238 volunteers on paper. Thirteen posted registry results and two are mentioned in the literature. Not one terminated trial is silent. The suspended one is: NCT06815471 — a CRF2 receptor agonist (COR-1167), investigated for worsening heart failure, with an estimated worldwide enrollment target of 300 participants. In the ClinicalTrials.gov record updated on 22 July 2026, recruitment was suspended on a DSMB recommendation to revise eligibility criteria. The target is not a confirmed enrollment count.

16 trials, 13,238 volunteers, ten international sponsors (Amgen, Corteria, Eli Lilly, GlaxoSmithKline, Johnson & Johnson, Novartis, Pfizer, Sanofi, Takeda, ViiV). Silence exists here too — in one case, suspended in 2026.

What you can do

05What you can do — patient, doctor, ethics committee

1. Before joining: ask for the trial’s public identifier: its EU CT number for a study registered in the European system, or its NCT number for one registered on ClinicalTrials.gov. Check the official record and ask the research team about authorisation, risks and consent before signing.
2. Study safety: ask who monitors adverse events and what to do if symptoms occur. Not every study has an independent DSMB/IDMC, and a serious adverse event does not automatically mean such a board has made a recommendation.
3. If a reaction occurs: contact the research team using the instructions provided and seek medical care when needed. In Romania, suspected medicine reactions can also be reported to ANMDMR; elsewhere in the EU, check your national authority. CTIS is the EU clinical-trials system, not a general patient reporting form.
4. About results: reporting duties and deadlines depend on the study, jurisdiction and applicable exceptions. Under FDAAA, the general deadline for applicable trials is one year after primary completion, subject to exceptions. A result not identified by this search does not by itself establish a violation.

What we don’t know

06What we don’t know — and why it is still enough

• “Mentioned, no results” (257): at least one paper cites the trial (MED source on EuropePMC), but the sponsor never posted results in the registry. Among suspended trials, the 16 mentioned ones are studies whose papers discuss the drug or the class (none was verified as a complete report); we did not count them as published.
• “Completely silent” (104): zero posted results in the registry and no mention of the NCT ID in the scientific literature (EuropePMC, 11 November 2026), and no paper identified in a second search (drug + condition + sponsor) read manually. For terminated trials, of 41 plausible secondary leads: 29 irrelevant, 9 almost certainly report the trial and 3 cannot be excluded (all also “mentioned”). For suspended ones, the 16 with appearances in the literature (15 with MED mentions + 1 lead that cannot be excluded) remain “mentioned”; the 16 with no trace at all remain silent.
• It is a register, not a conviction. A trial stopped for harm is often the most ethical outcome possible. What is missing is what comes after: the data never enters the literature, and other researchers re-run the same combination on other patients, without knowing.
• The categorisation rests on the sponsor’s stated reason. Some sponsors write “business decision” for what was in fact a safety problem; others write “safety” for commercial reasons. We left their reasons word for word — you can read each one.
• “Positive” stops do not appear here. Trials stopped because they worked (early efficacy signals) or for purely commercial reasons are not among the 624.
• Was this done before? Yes — the FDAAA reporting test (TrialsTracker, COMPare/Goldacre) has for years shown that a share of trials never post results on time. Nobody had published the complete register of those stopped for harm. This is it.

Method and sources

07Method, sources and disclaimers

What this page says: how many human trials marked “Terminated” or “Suspended” are listed on ClinicalTrials.gov; how many sponsor-stated reasons indicate harm; what became of their results — including the 104 completely silent ones.

What it does not say: that any of the 624 was illegal, immoral or unethical. Stopping or suspending for harm is often the best protection volunteers can get. What is missing is publication — and that is exactly what this page counts.

• ClinicalTrials.gov: NCT06815471 and FDAAA results reporting; EMA: reporting suspected side effects; European Commission: clinical trials and CTIS; NIH: study monitoring.
• ClinicalTrials.gov API v2, filter.overallStatus=TERMINATED (34,707 trials) and SUSPENDED (1,436), accessed 11 November 2026. Fields used: stopping reason, sponsor and sponsor class, phase, enrollment, start year, recruiting countries, posted results, published references.
• EuropePMC REST API, accessed 11 November 2026: search by NCT for all 624; a secondary search (drug + condition + sponsor) for the 169 with no posted results and no NCT mention; manual reading of every plausible secondary hit (41 leads for terminated trials, 12 reclassifications documented in the project data files; for suspended ones, 15 carry MED mentions and one (the FA + ropivacaine lead) cannot be excluded — all remain “mentioned”).
• The classifier and the manual reading passes: scripts/. All 1,938 reasons (1,826 terminated + 112 suspended) were read manually on 11 November 2026; the 624 confirmed and the 27 removed in the rescan are listed in the project data files (published alongside the page).
• Sponsors are named here by their official registered name. “Silent trials per sponsor” = the sponsor’s harm-motivated trials that posted no results and do not appear in the literature.
• This selection was reviewed only by its author — there is no external independent review (Codex CLI, usually used for this, is at its usage limit until 13 November 2026). The publication verdict is not a legal, medical or regulatory statement.
• Safety authorities: ANMDMR (Romania), EMA (EU), FDA (US, MedWatch). FDAAA 801 — results within 12 months; EU Regulation 536/2014 — traceability of trials in the EU.