Updated September 9, 2026: This piece was expanded with comments from researcher Nicholas DeVito on the comparability of disease-area compliance estimates, his assessment of the disclosure laws, and a new section on what disclosure timing doesn’t measure.
Updated September 13, 2026: A prior version of this piece described “five independent analyses” of FDAAA compliance. Two of the cited figures, the government-sponsor and industry-leader compliance rates, come from the same underlying cohort study as separate sponsor-type strata, not from two independent analyses. The piece has been corrected to reflect three independently conducted studies. No compliance percentage changed.
“Completed” is a status field, not a verdict. The distinction sounds obvious stated plainly, and it gets lost constantly in practice: investors scan a registry for a catalyst, clinicians decide what to trust, and a study showing COMPLETED next to a CNS or psychiatric indication with no publicly posted results gets read as a signal. This year the same state has recurred among unrelated sponsors, phases, and institutions. It is an information state, not an outcome, and the rules governing when that state should resolve are more specific, and more unevenly followed, than most coverage of clinical trial results assumes.
What “completed” actually promises, and what it doesn’t
ClinicalTrials.gov records two completion dates, and they do different work. The Primary Completion Date marks when the last participant finished data collection for the primary outcome; it is the date that starts the legal disclosure clock. The Study Completion Date, often later, marks the end of all data collection, including secondary and follow-up measures. Neither says anything about database lock, when the sponsor finishes cleaning and finalizing the data, nothing about topline results or a sponsor’s own summary announcement, and nothing about peer-reviewed publication, which routinely lags both by months or years. These steps are sequential, not simultaneous, and a reader who treats “completed” as shorthand for “known” is reading past three separate events that haven’t happened yet.
The rule, where one exists
Section 801 of the FDA Amendments Act of 2007 attaches a deadline to what the law calls Applicable Clinical Trials: generally Phase 2 through 4 studies of an FDA-regulated drug or biologic, and most device trials, with a U.S. site or conducted under an IND. Sponsors of those trials must post summary results to ClinicalTrials.gov within 12 months of the primary completion date. The deadline has lawful outs: a sponsor seeking initial approval of a new product, or approval of a new use, can certify for delayed submission, and good-cause extensions exist, so a trial past the 12-month mark is not automatically out of compliance.
The exclusions matter as much as the rule. Phase 1 drug and biologic trials are not covered. Small device feasibility studies are not covered. Purely observational studies are not covered. A meaningful share of completed-no-results trials sit in exactly this territory, with no statutory deadline attached at all: not evidence of anyone breaking a rule, because no rule applies.
Enforcement, where the rule does apply, has been close to theoretical. The statutory penalty, up to $10,000 per day of noncompliance at enactment and adjusted upward for inflation since, has essentially never been collected. Researcher Nicholas DeVito, who has spent years building public tooling to track this exact gap, put it to a reporter without much room for interpretation: to his knowledge, no fine has ever been issued.
“If you’re not going to wield any of the power given to you by the law, the law is effectively neutered.”
— Nicholas DeVito, researcher tracking clinical-trial reporting compliance
His assessment of the laws is more nuanced than the enforcement record alone might suggest. “These laws have, seemingly, helped us get some results out that otherwise wouldn’t be out there,” he said. “Is it perfect? Certainly not. Do we still have a problem specifically with late results? Definitely. But I’d much rather have these laws, even the watered down versions we have now where compliance is not rigorously enforced, than nothing at all.”
The posture, at least, changed this year. In March 2026, the FDA sent compliance reminders to more than 2,200 sponsors and researchers with overdue results. Whether those reminders translate into penalties remains unresolved.
The compliance data doesn’t converge on one number, and that’s the finding
Independent analyses of FDAAA compliance, spanning different time periods, sponsor types, and disease areas, return meaningfully different figures. One large cohort study found US government-sponsored trials had the lowest compliance of any sponsor class, at 31.4 percent within one year, while the largest individual pharmaceutical sponsors, tracked by that same research group as a separate stratum within the same cohort, posted 94 to 100 percent compliance. A separate, independent glioblastoma-specific cohort found 13 percent compliance within the 12-month deadline, rising to 82.7 percent within five years. A separate 2025 sponsor-level analysis put overall compliance at 37.2 percent and industry-only compliance at 73.7 percent, a figure that lines up closely with the FDA’s own internal estimate. That is three independently conducted studies, not four separate data points each standing on its own, and the range across even that more careful count still spans from 13 percent to 100 percent depending on sponsor type and disease area.
There is no single honest number to cite. Sponsor type predicts compliance more strongly than almost anything else measured, and disease area, trial phase, and time window all move the figure further. Any “trials are typically X percent late” claim would manufacture precision the underlying research doesn’t have. The heterogeneity is not a gap in the data; it is what the data shows.
DeVito points to another reason these numbers resist easy comparison: disease-area estimates can introduce methodological differences of their own. “I’ve not looked specifically at trials by any disease area really,” he said, “but at the same time, a bugaboo of mine is that they tend to have such different methodologies it’s difficult to make comparisons between them as to the true prevalence of non-reporting in an area.” The problem therefore extends beyond differences in sponsor type, phase, and observation window. Estimates drawn from different disease-area studies may also reflect differences in how those studies defined eligible trials, reporting, compliance, and follow-up, making apparent prevalence differences harder to interpret cleanly.
When the gap starts to mean something
Two situations produce genuinely different reads, and collapsing them into one is the most common interpretive mistake with this status.
A trial completed eight months ago, still inside the 12-month statutory window, with no results posted, is very likely routine. Nothing about that timeline should move anyone’s read of the underlying science.
A trial that has passed its required or expected disclosure deadline, whether the statutory window for an Applicable Clinical Trial after any certified delay, or a sponsor’s own previously stated milestone, and remains unposted on repeated checks has moved into a different category. Not evidence of a bad outcome. Evidence that the normal clock has run out without the expected event happening, which is worth tracking on its own, distinct from and prior to any conclusion about why.
What none of this licenses
No posted results does not mean a failed trial, and delayed publication does not mean concealment. Both readings require evidence beyond the absence itself. The accurate vocabulary is “results not yet publicly available,” “completed, results pending,” “disclosure gap.” Language implying hidden data or suppressed findings needs its own direct evidence before it appears in copy, not an inference from a blank field on a registry page.
Why this belongs on a watchlist, not just in a single article
A single completed-no-results trial tells a reader almost nothing on its own. The useful discipline is tracking the status across a portfolio: completed, results pending, results posted, published, discontinued, treated as a structured category rather than a fact rediscovered one trial at a time. A structured view of which trials are still pending, how long they have been pending, and whether that duration has crossed a known threshold tells a reader something closer to the actual maturity of the evidence base behind an indication or mechanism. That is what investors, clinicians, and researchers need to gauge.
What disclosure timing doesn’t measure
Disclosure, as the rules above define it, is a threshold question: were results posted, and within what window? DeVito points to a separate question sitting behind it. “Reporting is just the first, binary step,” he said. “We need to have all the results, full stop. However, at some point we also want to ensure that the results are reported fully, unbiased, and transparently so that we can appropriately assess their quality and incorporate them into the evidence base.”
The disclosure rules have comparatively little to say about that second problem: whether publicly available results provide a sufficiently complete and transparent account of what the trial found. DeVito calls this “an entire frontier of research in this area that is interesting and fruitful to consider.” A trial can meet its disclosure deadline and still leave open questions about selective reporting, completeness, and how faithfully the public record represents the underlying study. Those are different transparency questions from whether results appeared on time, and they require different tools to evaluate.
Why no single trial is named here
Every claim above rests on independently verified regulatory materials and peer-reviewed compliance research. The individual trials that first suggested the pattern are deliberately left out, because a disclosure status that cannot be confirmed cleanly across registry, sponsor, and phase should not be stated as fact. That difficulty is the point: disclosure status is harder to read precisely than a single registry field suggests.
Completed is where a trial’s data collection ends. It is not where a reader’s understanding of that trial should end.