ABOUT

Audit Clinical Evidence™ (ACE™) is an independent initiative that publishes detailed, reproducible audits of clinical and medical research. Each ACE report documents the errors, inconsistencies, and integrity concerns that can be verified from the public record of one or more published studies, and shows exactly how each finding was reached.

Mission

The purpose of ACE is to provide transparent, reproducible, and evidence-based audits of clinical and medical research. ACE reports focus on observable statistical errors, internal inconsistencies, image anomalies, potential data fabrication, publication ethics issues, and other research integrity concerns.

Clinical research influences meta-analyses, clinical guidelines, patient care, health policy, and public trust in medicine. When a published study contains impossible statistics, undisclosed protocol deviations, duplicated data, fabricated images, or other serious integrity problems, those problems propagate far beyond the original article. ACE exists to make such problems easier to identify, document, and act upon, by producing reports that journals, publishers, institutions, guideline developers, meta-analysts, clinicians, and readers can use to assess whether a study should be trusted as evidence.

The identification of errors or anomalies in ACE reports is strictly evidence-based. Findings are limited to issues that can be documented from the published record, trial registrations, supplementary materials, public databases, article metadata, and reproducible statistical checks. ACE avoids unsupported speculation, but may identify plausible integrity concerns when the public record contains strong convergent evidence.

Scope of ACE Reports

Each ACE report provides a structured audit of one or more related clinical research articles. Reports may examine randomized controlled trials, observational studies, meta-analyses, systematic reviews, clinical guidelines, or clusters of related publications.

Depending on the article(s) under review, an ACE audit may include checks for:

  • impossible or internally inconsistent summary statistics;
  • mismatches between reported test statistics, sample sizes, confidence intervals, and p values;
  • GRIM, GRIMMER, SPRITE, or related numerical consistency failures;
  • discrepancies between trial registrations and published outcomes;
  • duplicated, overlapping, or implausibly related datasets;
  • image or figure anomalies;
  • authorship, publication, or peer-review irregularities;
  • unexplained changes in sample size, analysis sets, or outcome reporting;
  • methodological errors that affect interpretation or downstream evidence synthesis.

When possible, ACE includes figures, tables, screenshots, code, and step-by-step explanations showing exactly where an error appears and how it was evaluated.

Evidence Standard

The goal of every ACE report is to separate what can be demonstrated from what can only be suspected. Every finding must satisfy three conditions:

  1. the issue must be observable in the public record (reported data, text, images, metadata, or registrations);
  2. the check must be reproducible by an independent reader;
  3. the conclusion drawn must not exceed what the available evidence reasonably supports.

These conditions ensure that each finding is tied to a transparent evidentiary basis. ACE reports do not accuse any author, institution, or publisher of deliberate misconduct; they document what the record shows and let readers draw their own conclusions.

Report Identifiers

Each ACE report receives a unique identifier made up of the project name, the publication year, and a three-digit report number.

For example, ACE2026001 is the first ACE report issued in 2026. The identifier appears on the report page, on the cover of the PDF, and in the report’s Zenodo record, so a report can always be located from its ID alone.

Versioning and Reproducibility

ACE reports are archived on Zenodo. Each published version of a report receives its own permanent DOI and is preserved as published, while the report’s concept DOI always resolves to the most recent version. Cite the concept DOI when referring to a report in general and the version DOI when a specific wording matters. Each report ends with a Version history section listing what changed in every revision, so readers can track how the report has evolved since it was first published.

Whenever possible, the code used to generate calculations, figures, tables, and consistency checks is included in the corresponding Quarto report files. The online report viewer includes collapsible code blocks so readers can inspect the code workflow without interrupting the main narrative. The PDF versions do not include code, in order to preserve readability and make the reports easier to share, cite, and review.

Corrections and Responses

Finalized ACE reports are routinely forwarded to the editorial boards and/or publishers of the corresponding journals for their independent review. Authors of an audited study who wish to provide data, clarifications, or a response are encouraged to contact ACE. By corresponding with ACE about a report, you agree that your correspondence may be published or quoted in part, with personal or sensitive information redacted, unless confidentiality is agreed in advance.

Readers may submit corrections to any report. Corrections are screened and, when credible evidence shows that a finding should be revised, incorporated into a new version of the report and recorded in its version history. See the disclaimers page for the full editorial and correspondence policy.

Suggesting a Study

If you come across a published study that looks problematic, you can write to ACE and suggest it. Readers, meta-analysts, editors, reviewers, and researchers who have noticed something in their own field are all welcome to do so.

If you would rather not be named, say so and ACE will not identify you publicly without your explicit permission. ACE will make all reasonable efforts to keep correspondence private, although privacy cannot be guaranteed against legal process.

Some concerns are not ACE’s to handle. If the matter appears criminal, or patients may be at risk, contact law enforcement or the relevant regulator. Formal misconduct allegations belong with the institution, funder, or research integrity body responsible for the work, which can compel access to data ACE cannot see. Writing to ACE is not a substitute for either.

Please do not send patient-identifiable data, or documents and datasets you are not permitted to share. What you send should rest on the published record and on material you are entitled to provide; ACE cannot accept or act on confidential internal documents.

A suggestion is most useful when it is specific. A general claim that a researcher, institution, journal, or public figure is “a fraud” is not something ACE can work from. What helps is naming the study, report, trial registration, dataset, figure, table, author group, journal, or publication record at issue and explaining what appears to be wrong. Useful things to include:

  1. the article title, DOI, journal, or trial registration number;
  2. the specific table, figure, outcome, image, statistic, or passage that appears problematic;
  3. what appears to be wrong, such as impossible statistics, duplicated data, image manipulation, authorship irregularities, undisclosed protocol deviations, plagiarism, or paper-mill indicators;
  4. any supporting documentation, screenshots, archived pages, PubPeer threads, registry records, code, calculations, or related publications;
  5. a brief explanation of why the issue matters.

ACE reads everything it receives, but writing to ACE does not guarantee that a report will follow. Because capacity is limited, priority goes to concerns that are specific, tractable, reproducible, and relevant to the reliability of clinical evidence.

Contact

Questions, corrections, suggestions for studies to audit, and responses from authors of audited studies can be sent to contact@auditclinicalevidence.org. If you would rather not be named, say so and ACE will not identify you publicly, as described above.