Corrections
We correct in the open. Every change to a published article is logged here, with a date.
The Vital Record treats corrections as first-class, public records. We do not silently edit published stories. When a factual error is found, the original text is preserved in the page’s history and a dated correction note is appended.
We use three levels:
- Clarification — wording improved; no factual change.
- Correction — a factual error fixed.
- Retraction — a claim that cannot be supported; the article stays online with a prominent retraction notice so the record is honest.
Log
Roche Terminates Phase 3 SKYSCRAPER-15 Tiragolumab Trial in Resected NSCLC, Deepening TIGIT Setbacks
Article described SKYSCRAPER-06 as a trial in 'advanced gastric or gastroesophageal junction cancer'; multiple primary sources (PubMed DOI:10.1001/jamaoncol.2024.4905, Lancet Oncology, ASCO Post) confirm SKYSCRAPER-06 (NCT04294810) was a Phase 3 trial in first-line advanced non-squamous NSCLC comparing tiragolumab plus atezolizumab and chemotherapy versus pembrolizumab and chemotherapy; indication corrected.
Original title stated 'While Adults Stay Protected'; Cao lab preprint (DOI:10.64898/2026.06.05.730251) limits adult protection to those with ancestral-strain SARS-CoV-2 imprinting—adults exposed only to Omicron subvariants are described as 'weakly imprinted' and would not retain equivalent protection; title qualified to specify ancestral exposure.
DEK stated 'eight months after a Refuse-to-File stumble'; the article body itself records the RTF date as February 3, 2026, which is approximately four months before the June 18 VRBPAC meeting—not eight; corrected to 'four months'.
Investigational Once-Weekly HIV Pill Meets Primary Endpoints in Both Phase 3 ISLEND Trials
Article described the drug as 'Merck's islatravir-based regimen' without naming lenacapavir (Gilead's capsid inhibitor co-drug); ClinicalTrials.gov (NCT06630286, NCT06630299) titles the regimen 'Islatravir/Lenacapavir' and lists Gilead Sciences as primary sponsor; article updated to name both drugs and both companies.
Aficamten Beats Metoprolol on Exercise Capacity in Phase 3 Head-to-Head Trial for Obstructive HCM
DEK and body claimed superiority 'across every/all key pre-specified secondary endpoints'; NEJM primary paper (DOI:10.1056/NEJMoa2504654) explicitly states left ventricular mass index—a registered secondary endpoint—did not differ significantly between groups; corrected to five of six pre-specified secondary endpoints.
A First-Line Challenger: What MAPLE-HCM Means for the Clinic
SEQUOIA-HCM described as a '28-week placebo-controlled trial'; ClinicalTrials.gov NCT05186818 and NEJM DOI:10.1056/NEJMoa2401424 both confirm the double-blind treatment period was 24 weeks; corrected to 24-week.
Original headline read 'Nine-State Cluster'; CDC and FDA primary sources confirm confirmed cases span three states (Maryland, New York, Virginia) and the recall distribution covers five states plus DC—no source supports a nine-state figure; headline corrected to 'Three-State Outbreak'.
Article stated 'not found in the United States since 1966' and 'first re-emergence since 1966'; New World Screwworm re-emerged in the Florida Keys from September 2016 through March 2017, killing approximately 135 endangered Key deer before eradication; headline, dek, and body corrected to specify 'mainland' and acknowledge the 2016–17 Florida Keys episode.
Article grouped Candida albicans under 'bacteremic pathogens'; Candida albicans is a fungus, not a bacterium—the correct clinical terms are candidemia or fungemia, not bacteremia; collective term corrected to 'bloodstream pathogens'.
A leaderboard-topping LLM still botches potassium dosing in a preprint
An earlier headline and lede called GPT-5-Chat “leaderboard-topping.” The preprint describes it as a model that appears on the MedAgentBench leaderboard, not one that tops it — that benchmark is in fact led by other models. The wording was changed to “leaderboard-listed.” Flagged by the post-publication verification pass.
FDA labels BD Pyxis cabinet recall Class I after fingerprint-scanner fault
An earlier version described the recall as Class I, the FDA's most serious tier. The FDA recall database lists both records (Z-2298-2026 and Z-2297-2026) as Class II; the Class I label applies to the cabinet's device regulatory category, not to this recall. The headline, standfirst, chart and text were corrected, and the November 2024 warning-letter figures were attributed to MedTech Dive's reporting. Flagged by the post-publication verification pass.