Only about 1 in 4 completed U.S. clinical trials post results — and industry sponsors report roughly twice as often as academic/hospital sponsors

Summary

Only a minority of registered clinical trials ever finish and publicly post results. In a current cross-section of 289,772 U.S. interventional trials in the AskAmerica warehouse, 47.4% have reached COMPLETED status, and of those, only 25.4% have posted results to ClinicalTrials.gov's results database. Reporting varies enormously by sponsor: federal-agency and NIH-funded completed trials post results 54-63% of the time, industry-funded trials about 43% of the time, and the large 'OTHER' category — academic medical centers, hospitals, foundations, and individual investigators, which together account for 70% of all completed trials — posts results only 18.6% of the time. This sponsor ordering (federal/NIH ≥ industry >> academic/other) matches the peer-reviewed literature (DeVito et al., The Lancet, 2020; Anderson et al., NEJM, 2015), though the exact percentages differ because those studies track prospective cohorts anchored to each trial's legal reporting deadline, while this analysis is a present-day snapshot of ClinicalTrials.gov's current registry.

What 'finish and report' means here, and why it matters

ClinicalTrials.gov distinguishes two separate things that are easy to conflate: overall_status = COMPLETED (the trial finished) and has_results = true (the sponsor posted a structured results record to the registry's results database). A trial can be completed for years with zero results ever posted — that gap is exactly what the 2007 FDA Amendments Act (FDAAA) and its 2017 Final Rule were written to close, by legally requiring many trials to report within 12 months of their primary completion date.

Using the AskAmerica health.clinical_trials table (one row per NCT ID, restricted by ETL design to studies with a ClinicalTrials.gov last-update-post-date on or after 2020-01-01 — i.e., trials still active in the registry's recent record, not a fixed historical cohort):

• Of 289,772 interventional trials, 137,279 (47.4%) have reached COMPLETED status. The rest are recruiting, active-not-recruiting, terminated, withdrawn, or of unknown status.
• Of the 137,279 completed trials, only 34,918 (25.4%) have has_results = true.

This 25.4% figure should not be read as 'the' national compliance rate for trials that are legally required to report (FDAAA only applies to a subset of trials, called Applicable Clinical Trials, mostly drug/device interventional trials past a certain phase) — it is the reporting rate across all completed interventional trials in the current registry, most of which are not legally obligated to report at all. It is a useful, computable proxy for the scale of the transparency gap, not a legal-compliance statistic.

How it varies by sponsor (funder_type)

Two things drive the low national aggregate (25.4%): (1) the OTHER category — non-NIH academic institutions, hospitals, foundations, and individual investigator-sponsors — is both the single largest sponsor class (69.6% of all completed trials, computed via subgroup_contribution) and has the lowest reporting rate (18.6%); and (2) OTHER_GOV (state and local government sponsors, a much smaller group at 2,986 completed trials) reports results only 3.7% of the time, the worst rate of any group, though its small share of the total limits its effect on the aggregate. Industry, despite running some of the largest and most scrutinized trials, still only reports about 43% of the time in this snapshot — well below the 70%+ compliance figures often cited for FDAAA-covered, legally-obligated industry trials specifically, because this figure spans ALL industry-funded interventional trials, not just the legally-covered subset.

How this compares to the peer-reviewed literature

The definitive academic source on FDAAA compliance is DeVito, Bacon & Goldacre, 'Compliance with legal requirement to report clinical trial results on ClinicalTrials.gov: a cohort study' (The Lancet, 2020), built from the Oxford Bennett Institute's FDAAA TrialsTracker (fdaaa.trialstracker.net). That study, which tracks trials prospectively from their legal reporting deadline rather than taking a cross-sectional snapshot, found that compliance with FDAAA was poor and not improving over its study window, and that non-industry sponsors (universities, hospitals, government) were far more likely to miss the legal deadline than industry sponsors, with U.S. government-sponsored trials the least likely to report on time. An earlier study, Anderson et al. (NEJM, 2015), found that only about 49-50% of NIH-funded trials and 42-45% of trials funded by other government or academic institutions had either reported results or obtained a legally acceptable delay, versus higher rates for industry.

The sponsor ordering in this warehouse's current snapshot (federal/NIH highest, industry next, academic/other-institution sponsors lowest) is directionally consistent with both of these peer-reviewed cohort studies. The exact percentages differ for a methodological reason worth stating plainly: the Lancet and NEJM studies use prospective cohorts of trials tracked from their FDAAA-defined due date and restrict to legally Applicable Clinical Trials, while this analysis is a simple cross-sectional snapshot of the current ClinicalTrials.gov registry across all interventional trials regardless of legal reporting obligation. The two approaches answer related but not identical questions, and no rate from one design should be read as a discrepancy with the other.

Source-reliability note: several web search results returned a specific set of sponsor-level compliance figures (73.7% industry, 71.0% NIH, 25.5% academic) attributed to a site called publichealth.realclearjournals.org. That domain returned an HTTP 403 error on direct fetch and is not a recognized peer-reviewed or agency source; consistent with this harness's own prior finding that this exact domain surfaces on this exact topic and is not independently verifiable, those specific figures are excluded from this report and are not used anywhere above.

Every query behind this report

5 warehouse calls ran in this session, in order. Each is reproducible against the same snapshot.

SELECT overall_status, COUNT(*) AS n, SUM(CASE WHEN has_results THEN 1 ELSE 0 END) AS n_results
FROM health.clinical_trials
WHERE type = 'trial'
GROUP BY overall_status
ORDER BY n DESC

SELECT type, COUNT(*) FROM health.clinical_trials GROUP BY type

SELECT overall_status, COUNT(*) AS n, SUM(CASE WHEN has_results THEN 1 ELSE 0 END) AS n_results
FROM health.clinical_trials
GROUP BY overall_status
ORDER BY n DESC

SELECT funder_type,
  COUNT(*) AS n_total,
  SUM(CASE WHEN overall_status = 'COMPLETED' THEN 1 ELSE 0 END) AS n_completed,
  SUM(CASE WHEN has_results THEN 1 ELSE 0 END) AS n_results,
  SUM(CASE WHEN overall_status = 'COMPLETED' AND has_results THEN 1 ELSE 0 END) AS n_completed_with_results
FROM health.clinical_trials
WHERE study_type = 'INTERVENTIONAL'
GROUP BY funder_type
ORDER BY n_total DESC

SELECT funder_type, 1 AS one FROM health.clinical_trials WHERE study_type='INTERVENTIONAL' AND overall_status='COMPLETED'