Artificial Intelligence replication of District Hospital Emergency Department Surge Staffing Under Unmeasured Triage Nurse Absenteeism: A Healthcare Discrete-Event Robustness Evaluation

Triage nurse absenteeism in district hospital emergency departments is rarely recorded at the moment it matters, so surge staffing models that condition on observed absenteeism are estimated against a covariate that is systematically missing precisely when the department is under strain. the present analysis treats the problem as one of estimator robustness rather than imputation. A discrete-event replication of a district emergency department is coupled to a linear regression of patient waiting time on a surge exposure, a staffing-group indicator and a patient-severity score, and the ordinary least squares estimator is characterised across 600 replications of N = 500 patients drawn from a declared data-generating process with intercept 12.0, exposure 0.42, group -3.1, severity 1.35 and error standard deviation 4.2. The estimated bias was 0.07133 for the intercept and -0.0012 for the exposure coefficient; the root mean squared errors were 0.84304 and 0.01545 respectively; and the empirical coverage of the nominal 95% confidence intervals was 1.0 for both parameters. The exposure coefficient was therefore recovered with negligible bias and near-nominal interval behaviour under the stated process, which establishes a benchmark against which absenteeism-contaminated estimators can be judged. The contribution is a reproducible robustness baseline, not a claim that unmeasured absenteeism is harmless: the replication isolates the cost of the missing covariate by holding the process fixed and reporting what the uncontaminated estimator does before the contamination is introduced.