Emergency Department Patient‑Flow and Provider‑Workflow Outcomes after Electronic Health Record Implementation or Optimisation: A Systematic Review

Abstract

Background: Electronic Health Records (EHRs) are embedded in every step of Emergency Department (ED) care, triage, medication administration, documentation and handover, but their effects on patient flow and clinician workload vary. In a crowded ED, even small delays in these steps matter for safety and access to beds. Decision-makers need clear evidence on whether to replace a whole system or improve targeted parts, and what to expect during the change. Aim: To estimate the impact of EHR implementation or major optimisation in hospital EDs on objective patient-flow and provider-workflow outcomes, and to identify contextual factors that modify these effects. Methods: MEDLINE (via PubMed), Scopus, and CINAHL were searched for English-language studies (January 2013–April 2025). Screening followed the PRISMA 2020; risk of bias was assessed with ROBINS-I, and results were narratively synthesised using Synthesis Without Meta-analysis (SWiM) (no metaanalysis). Ten studies met the inclusion criteria. Results: In the included studies, Targeted EHR upgrades, such as focused care pathways, acuity-based assignment and structured handover, most often improved flow, typically cutting wait times by tens of minutes. In contrast, “big-bang” system replacements commonly produced a short, predictable dip in ED performance, with length of stay and door-to-provider times rising initially and trending back toward baseline within about six months. Documentation time increased right after go-live, but often lessened as voice recognition, templates and shortcuts matured. Clinician throughput dipped at go-live and partially or fully recovered when efficiency tools were adopted. Confidence in effect sizes was limited by confounding and incomplete reporting of implementation fidelity. Conclusion: EHR changes behave like complex interventions with a learning curve, disruption, stabilisation, and then optimisation. Targeted modules can reliably help flow; whole-suite go-lives need temporary buffers, super-users and rapid feedback. Future evaluations should use robust quasi-experimental designs and clearly defined ED measures tracked at regular intervals around implementation.