Case study: a worked review of corrected PREDIMED

This example compares the concerns in a RigorMD report with the randomization deviations disclosed in the corrected PREDIMED paper. The report quotes the manuscript as evidence for its findings. Those quotations show where the concern appears; they do not establish that the models' knowledge of the trial's history played no role.

The first case is PREDIMED — Estruch et al., a landmark trial of a Mediterranean diet for cardiovascular prevention. The original 2013 report (N Engl J Med 2013;368:1279-90) ↗ was retracted and republished in 2018 after questions about its randomization.

The trigger was statistical detective work. John Carlisle's survey of baseline data in 5,087 randomized trials ↗ flagged trials whose baseline distributions were too similar or too divergent to be consistent with true random sampling — and PREDIMED was among the papers this method drew attention to. On investigation, the authors confirmed the concern: randomization had not been done properly for a substantial minority of participants. The journal issued a retraction-and-republication notice ↗ — “Because of irregularities in the randomization procedures…” — and published a reanalysis (N Engl J Med 2018;378:e34) ↗.

Appraising the 2018 republished article — the current authoritative version, which itself documents the deviations — the engine's top finding was compromised randomization for roughly 1,588 of 7,447 participants (about 21%): household members assigned to a relative's arm, one site randomizing by clinic rather than by individual, another applying its randomization tables inconsistently. For that subset, the trial is effectively quasi-randomized, and the effect there is identified by propensity adjustment rather than randomization — so unmeasured confounding cannot be excluded.

That finding is the same issue as the documented retraction reason. Both engines flagged it independently, and the deterministic layer separately reconciled the reported hazard ratio with its interval and p-value and confirmed the event counts, with no inconsistencies. The concordance is on the primary methodological problem — the thing the record already knew. Randomization integrity is the CONSORT question worked through in our CONSORT walkthrough.

Catching the issue is only half the test. The engine rated it moderate — not serious or critical — because the republished paper discloses every deviation and shows the protective effect survives propensity-adjusted reanalysis and the omission of the 1,588 affected participants. This is the grade observed for this case; it does not establish calibration across studies. The NASCIS case shows a different reported concern: a conclusion resting on a subgroup while the primary comparison was null is graded serious, and the paper's confident advice about complete injuries, which reaches past a small subgroup and is fixed by rewording, is graded moderate.

The models may know the paper and its later history. This review cannot establish what they would have found at the original submission or measure sensitivity, specificity, or accuracy.

The validation page lays out the claim map, the matched quotes, the reported domain scorecard, and the honesty notes on which version was appraised and why.

Read the worked review of the corrected PREDIMED paper →, see how the engine works, or read a sample report and review pricing — the pre-submission review is $99.