Shouldn't we want replication even if the sample size seems large, to rule out experimental errors or other flaws in the work?

It isn't the most diverse cohort so a body such as the WHO might or might not choose to make recommendations based on this single finding.