Abstract / Summary
Hospitals have limited capacity to review discharge medication regimens. This planned retrospective, single-center study will use linked MIMIC-IV and MIMIC-IV-Note records to evaluate whether contextual information improves the prioritization of opioid prescribing and reconciliation concerns after orthopedic surgery, compared with dose-based models and transparent clinical rules. The study connects clinical epidemiology, biostatistics, and health-services research through three aims: establish the validity of the cohort and medication measurements; compare clinically confirmed review yield at a fixed review capacity against blinded, independent clinical adjudication; and assess documentation gaps, unassessable records, and subgroup differences in coverage and performance. Reviewer workload and time will be evaluated where measured. An offline clinician-reader experiment is optional. The intended public-health contribution is evidence to inform the allocation of scarce medication-review resources while protecting appropriate pain treatment. Equity analyses will examine who may be excluded, missed, or subjected to unnecessary review because of data limitations or model errors, with conclusions constrained by sample size and measurement quality. The system is intended to support human review, not autonomous prescribing or automatic opioid reduction. This is a research protocol, not a completed validation or a report of study results. No credentialed patient data were accessed or analyzed in preparing the attached package. The study will not establish reductions in overdose, opioid consumption, or population-level harm; policy effectiveness, cost-effectiveness, and transferability beyond the source setting require subsequent evaluation.