Machine Learning Triage of Emergency Department Imaging Under Unmeasured Radiologist Queue-Jumping Authority: A Healthcare Governance Difference-in-Differences Evaluation

Machine learning triage systems for emergency department imaging are typically evaluated as if the radiologist's queue were a neutral ordering device, yet senior radiologists routinely exercise unmeasured queue-jumping authority that reorders the worklist before any algorithm can act on it. the present analysis treats that authority as a governance variable rather than a nuisance and asks what a before-after-control-impact design can and cannot identify about triage effects when queue-jumping is unrecorded. Because no observational data are available, the study characterises the operating characteristics of the design under a declared variance structure: between-cluster standard deviation 6.0, residual standard deviation 4.5, and intraclass correlation 0.64. Power was estimated across 2000 design cells spanning cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. The analysis indicates that 54 clusters are required to detect a 5 percentage point effect at 80 per cent power under this variance structure, a requirement that exceeds the largest cluster count examined. The contribution is a governance-facing account of what difference-in-differences can credibly attribute to triage algorithms when radiologist discretion remains outside the measurement frame, and why the design's power ceiling is itself a political fact about departmental authority.