Which randomized trials of ivermectin were later retracted or found to have integrity problems, and how did that affect meta‑analyses?
This fact-check may be outdated. Consider refreshing it to get the most current information.
Executive summary
A small set of randomized trials of ivermectin for COVID‑19 were later retracted or flagged for serious integrity problems — most prominently the Elgazzar trial and several other studies with impossible numbers or randomization failures — which materially distorted early meta‑analyses that showed large benefits; when those suspect trials were removed or stratified by quality, apparent survival and hospitalization benefits largely disappeared [1] [2] [3]. Major reviews and journal editors concluded that the evidence base was unreliable and that some high‑profile meta‑analytic claims depended on data later judged fraudulent or at high risk of bias [4] [5].
1. Which randomized trials were later retracted or flagged for fraud or grave methodological problems
The most widely discussed single trial was the Elgazzar study — a large randomized trial that was withdrawn after allegations of fraudulent data and subsequently led authors of a prominent meta‑analysis to retract their paper because that trial had been a major weight in pooled estimates [1] [4]. Other trials were singled out for problems suggestive of randomization failure or impossible numbers; for example, Niaee et al. was questioned for baseline imbalances and unusual timelines, and independent reviewers flagged at least two trials with stratification or numerical anomalies that implied severe methodological concerns [2] [3] [6]. Systematic checks across 26 major trials found that over one third had “serious errors or signs of potential fraud,” according to reporting in The BMJ and related reviews [5].
2. How those problem trials changed early meta‑analytic conclusions
Early meta‑analyses that pooled many small trials — some unpublished or on preprint servers — produced optimistic pooled effects largely because a few large or apparently favorable trials contributed disproportionate influence; Nature Medicine noted that a single flawed randomized trial represented more than 10% of the overall effect in at least two major meta‑analyses [3]. Hill and colleagues published a meta‑analysis that initially suggested benefit but retracted it after learning one of the largest included trials was withdrawn for fraudulent data, then reworked the analysis stratifying by trial quality [7] [2]. Re‑analyses that excluded retracted or high‑risk studies generally lost the statistically significant survival or hospitalization advantages previously reported, and editorial expressions of concern warned that excluding suspicious data “appears to invalidate” earlier claims of mortality reductions [8] [4] [5].
3. Specific quantitative impact on pooled estimates and certainty
The retraction of the single large fraudulent trial and removal of other suspect studies shifted pooled point estimates toward no effect and widened confidence intervals; in the retracted Hill meta‑analysis, when higher‑risk studies were excluded the primary analysis showed no statistically significant difference in survival or hospitalization (risk ratio for survival 0.90, 95% CI 0.57–1.42; RR for hospitalization 0.63, 95% CI 0.36–1.11) — findings the authors later corrected and republished after addressing bias and suspected fraud [8] [2]. Multiple later systematic reviews that used stricter inclusion and bias‑stratification reached the conclusion that the best available randomized evidence does not support a clear mortality or hospitalization benefit and that earlier positive signals were driven by low‑quality or fraudulent studies [6] [9].
4. Broader implications, dissenting analyses, and remaining uncertainties
The ivermectin episode illustrated how meta‑analyses that rely on summary data and poorly vetted, rapidly posted trials can be misled by a handful of contaminated datasets, prompting calls for individual patient data checks and higher standards for living reviews [3] [10]. Some pro‑ivermectin aggregators continue to publish favorable pooled results after disputing exclusions and assessing different quality criteria, but major journals and independent reviews have emphasized that including retracted or clearly biased trials inflates effect sizes and undermines guideline‑level conclusions [11] [12]. Remaining uncertainty stems from uneven trial quality and from the fact that definitive large, well‑conducted phase‑3 randomized trials were still ongoing at the time of these reports; reviewers therefore caution that claims of benefit based on the tainted early literature are not reliable [13] [14].