Check 1 (Verbatim Priority): The verbatim quote is assessed: 'Three passengers have died. Three others who are sick were evacuated today, and two others are hospitalized, including one in Switzerland.' Check 2 (Extraction Fidelity): The claim_text adds 'passengers' to the evacuee characterization, but the verbatim does not — the evacuees are called 'three others,' not 'three passengers.' This extraction-introduced language is noted but does not affect the verdict, as the verbatim is the operative text. Sub-claim A (three deaths): Fully confirmed by multiple Tier 1-2 sources including the WHO Director-General's May 7 briefing and the ECDC assessment, both of which document three deaths linked to the outbreak — a Dutch couple and a German national. Sub-claim B (three others who are sick were evacuated): Three people were indeed evacuated on May 6, confirmed by WHO, ECDC, CBS, NBC, Al Jazeera, NPR, and the cruise operator. However, the verbatim characterizes all three as 'sick.' This is an identifiable inaccuracy: WHO and cruise operator Oceanwide Expeditions both stated that of the three evacuees, two were symptomatic and one was asymptomatic — evacuated as a precautionary measure due to close contact with a confirmed case. Describing an asymptomatic individual as 'sick' is a specific factual element that is contradicted by primary institutional sources. This does not reverse the directional meaning (three were evacuated due to the outbreak), making MOSTLY TRUE the appropriate verdict rather than FALSE (per MOSTLY TRUE boundary test: core assertion substantially correct with one identifiable inaccuracy that does not reverse directional meaning). Sub-claim C (two others hospitalized, including one in Switzerland): This is confirmed. A British national was in ICU at a Johannesburg private facility confirmed as hantavirus-positive, and a Swiss male former passenger was being treated at University Hospital Zurich — both documented by WHO, ECDC, CNN, CBS, and NBC as of May 6, the air date. The claim does not overcount or misplace the hospitalized cases. The single identifiable inaccuracy — 'sick' applied to an asymptomatic evacuee — is the basis for MOSTLY TRUE rather than TRUE, consistent with the methodology's immaterial imprecision standards: while approximate language is tolerated, a specific outcome characterization ('sick') directly contradicted by the primary institutional record on a named sub-element is not merely immaterial.

Methodology note: The TRUE/MOSTLY TRUE boundary here turns on whether the word 'sick' applied to one asymptomatic evacuee constitutes an immaterial imprecision or an identifiable inaccuracy under the methodology. The methodology's immaterial imprecision standards specify that 'specific outcome claims' unqualified by context and 'contradicted by the primary institutional record' are MOSTLY TRUE, not TRUE. WHO and the cruise operator both affirmatively stated the evacuee was asymptomatic, making this a record-contradicted specific characterization rather than approximate language. The MOSTLY TRUE nomination is therefore consistent with the methodology even though the inaccuracy is narrow. Reviewers should note that had Cooper used a neutral term (e.g., 'three others were evacuated'), the verdict would be TRUE.