Methodological Analysis of Bias Risks in Adaptive Multi-Arm Platform Trials: A Case-Series from Three COVID-19 Studies

Reich, S., Center for Open Science, doi:10.31222/osf.io/h5kc8_v1, Jan 2026
Ivermectin for COVID-19
4th treatment shown to reduce risk in August 2020, now with p < 0.00000000001 from 106 studies, recognized in 24 countries.
No treatment is 100% effective. Protocols combine treatments.
6,300+ studies for 210+ treatments. c19early.org
Review of methodological biases in three major adaptive platform trials (ACTIV-6, PRINCIPLE, and TOGETHER) evaluating ivermectin for COVID-19. Author finds that these influential studies were compromised by extensive post-enrollment protocol amendments, suboptimal dosing, and reporting inconsistencies that systematically biased results toward underestimating efficacy. Author details how primary endpoints were frequently modified after enrollment had begun, while eligibility criteria were expanded to include lower-risk populations, effectively reducing the power to detect differences in severe outcomes. Author also highlights that the dosing regimens were often below pharmacokinetic recommendations and, crucially, administered on an empty stomach, which significantly reduces bioavailability compared to administration with a high-fat meal. Delayed treatment initiation, with patients often treated several days after symptom onset, was identified as further limiting potential antiviral benefits. Author also notes that positive findings in secondary outcomes were often downplayed or omitted in final publications, and that transparency was lacking due to the failure to share de-identified participant data despite registry promises. Details of all issues reported, and many more, can be found in the individual paper analyses1-3.
Reviews covering ivermectin for COVID-19 include4-52.
Reich et al., 6 Jan 2026, preprint, 1 author.
DOI record: { "DOI": "10.31222/osf.io/h5kc8_v1", "URL": "http://dx.doi.org/10.31222/osf.io/h5kc8_v1", "abstract": "<p>Background: Adaptive platform trials (APTs) were widely promoted during the COVID-19 pandemic as an efficient and flexible approach for evaluating multiple interventions. However, their complexity and frequent protocol amendments may introduce risks of bias. We selected ivermectin as our case study both because of the marked discrepancy between the generally positive findings in the broader literature and the negative results from a few large trials, and because these influential studies were all designed as adaptive platform trials with multiple arms.Methods: We performed a methodological case-series analysis of three major adaptive platform trials: ACTIV-6 (USA), PRINCIPLE (UK), and TOGETHER (Brazil), each of which included an ivermectin arm for outpatient COVID-19. For each trial, we systematically reviewed all publicly available protocols, registry entries, analysis plans, preprints, and publications. We identified key protocol changes, design limitations, and reporting issues, and summarized recurring patterns related to endpoints, dosing, treatment timing, randomization, reporting, and data transparency. Each issue was assessed for its likely direction of bias on the observed efficacy of ivermectin. Results: All three trials exhibited extensive protocol changes, outcome modifications, and reporting inconsistencies, many of which occurred after participant enrollment had begun. Common issues included delayed treatment initiation, suboptimal dosing or administration, selective reporting of results, and inconsistent application of early stopping rules. Positive findings in secondary outcomes or subgroups were often omitted or downplayed in final publications. Data inconsistencies and lack of transparency further limited independent verification. The cumulative effect of these issues was a systematic bias toward underestimating any potential benefit of ivermectin.Conclusion: Our analysis shows that even large adaptive platform trials can be compromised by methodological flaws, losing their intended advantages and potentially distorting the evidence base. In all cases, these flaws systematically biased the assessment of ivermectin, meaning the results of these trials cannot be regarded as definitive evidence against its potential benefit. We propose that small but well-designed randomized trials, when combined in meta-analyses, may provide a more reliable foundation for evidence, even if individual studies are underpowered. Above all, critical appraisal and transparency are essential for trustworthy clinical research and sound policy decisions.</p>", "author": [ { "affiliation": [], "family": "Reich", "given": "Simon", "sequence": "first", "suffix": "MD" } ], "container-title": [], "content-domain": { "crossmark-restriction": false, "domain": [] }, "created": { "date-parts": [ [ 2026, 1, 6 ] ], "date-time": "2026-01-06T18:59:43Z", "timestamp": 1767725983000 }, "deposited": { "date-parts": [ [ 2026, 1, 7 ] ], "date-time": "2026-01-07T05:01:49Z", "timestamp": 1767762109000 }, "group-title": "MetaArXiv", "indexed": { "date-parts": [ [ 2026, 1, 7 ] ], "date-time": "2026-01-07T05:24:18Z", "timestamp": 1767763458130, "version": "3.48.0" }, "is-referenced-by-count": 0, "issued": { "date-parts": [ [ 2026, 1, 6 ] ] }, "license": [ { "URL": "https://creativecommons.org/licenses/by-nd/4.0/legalcode", "content-version": "unspecified", "delay-in-days": 0, "start": { "date-parts": [ [ 2026, 1, 6 ] ], "date-time": "2026-01-06T00:00:00Z", "timestamp": 1767657600000 } } ], "member": "15934", "original-title": [], "posted": { "date-parts": [ [ 2026, 1, 6 ] ] }, "prefix": "10.31222", "published": { "date-parts": [ [ 2026, 1, 6 ] ] }, "publisher": "Center for Open Science", "reference-count": 0, "references-count": 0, "relation": {}, "resource": { "primary": { "URL": "https://osf.io/h5kc8_v1" } }, "score": 1, "short-title": [], "source": "Crossref", "subject": [], "subtitle": [], "subtype": "preprint", "title": "Methodological Analysis of Bias Risks in Adaptive Multi-Arm Platform Trials: A Case-Series from Three COVID-19 Studies", "type": "posted-content" }
Please send us corrections, updates, or comments. c19early involves the extraction of 200,000+ datapoints from thousands of papers. Community updates help ensure high accuracy. Treatments and other interventions are complementary. All practical, effective, and safe means should be used based on risk/benefit analysis. No treatment or intervention is 100% available and effective for all current and future variants. We do not provide medical advice. Before taking any medication, consult a qualified physician who can provide personalized advice and details of risks and benefits based on your medical history and situation. IMA and WCH provide treatment protocols.
  or use drag and drop   
Submit