Answers are grounded in this paper and cite their sources.
Background/Objectives: Emergency department (ED) demand models often rely on time-of-day and calendar patterns, yet accurate models do not automatically change practice. We examined what a predictive model adds to the judgement of clinicians, operational managers, and analysts, and what affects whether its signal is used. Methods: We used a convergent mixed-methods design within the evaluation phase of a design-science programme in one district of Aotearoa New Zealand. Seven semi-structured interviews with clinical, operational, and analytical staff were analysed using reflexive thematic analysis with hybrid coding; the model was withheld until the second half of each interview, so participants first described demand from experience. Routine ED arrival data for 2017–2023 from the same district were analysed descriptively as unadjusted daily counts, and the strands were integrated in a joint display interpreted through the Theory of Swift, Even Flow (TSEF) and Normalisation Process Theory (NPT). Acuity was approximated by arrival by emergency transport, which marks resource intensity at arrival rather than clinical severity. Results: Thirty-seven codes formed four themes. Before seeing the model, participants in all role groups described the main demand patterns, including daily and weekly cycles, winter pressure, and doubts about event-related surges, and the descriptive data were consistent with these views; in unadjusted comparisons, event days were no busier than other days (342.7 vs. 351.4 mean daily presentations). Participants judged that the model mainly formalised what they already knew and located its potential value in a clear, timely signal that shows demand against a normal baseline and links to inpatient discharge as well as staffing, with equity monitoring and Māori data sovereignty built in from the start. Participants, especially those in operational roles, placed the main constraint in access block, which the arrival data did not measure. Conclusions: We propose predictive redundancy as an interpretive concept for this overlap: for the predictable, calendar-driven component of daily arrivals, a model may add little to what experienced staff already know, so its value shifts from discovery to standardisation. Forecast accuracy was not compared with expert forecasts, so this proposition needs quantitative testing. TSEF identifies what a forecast should target, and NPT explains whether staff will use it.
No takes yet. Share an insight, caveat, or question.
PoshtMashhadi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: