The "RCT reproducibity problem" has a darker and harmful relative.... "unsafe transport of trial entry criteria and treatment to clincial protocols". The COVID-19 ventilator harm experience exposed a major RCT design flaw at scale: During the pandemic early ventilator protocols calling for High PEEP tables was rapidly promulgated as "strong" and RCT meta-analysis evidence based, but instead produced harm at scale and were abandoned by bedside clinicians. This sentinal event cannot be ignored and requires deep epistemic investigation because it demontrates that flawed RCT design can pose a public safety risk at scale. This review demonstrates that that many decades ago a new species of RCT emerged which is structurally unsafe because they test interventions in cause-agnostic, syndrome-defined populations and then export those results into bedside protocols as if they were disease-specific truths. By treating consensus thresholds (e.g., ARDS, sepsis, CAP) as transportable diagnostics rather than triage labels, these trials achieve internal validity while lacking causal alignment, creating a predictable failure mode when protocols are applied to new diseases or contexts.