The effectiveness of point of care high sensitivity troponin testing to improve Emergency Department flow: a multi-centre controlled interrupted time series

@misc{mchenry2026effectiveness,
  title = {{The effectiveness of point of care high sensitivity troponin testing to improve Emergency Department flow: a multi-centre controlled interrupted time series}},
  author = {R. D. McHenry and A. Saunders and F. Ahmad and D. Mackay},
  howpublished = {medRxiv},
  year = {2026},
  month = aug,
  doi = {10.64898/2026.08.27.26361548},
  url = {https://www.medrxiv.org/content/10.64898/2026.08.27.26361548v1},
  note = {CC BY-NC},
}

TY  - UNPB
TI  - The effectiveness of point of care high sensitivity troponin testing to improve Emergency Department flow: a multi-centre controlled interrupted time series
AU  - R. D. McHenry
AU  - A. Saunders
AU  - F. Ahmad
AU  - D. Mackay
PY  - 2026
DA  - 2026/08/27
PB  - medRxiv
DO  - 10.64898/2026.08.27.26361548
UR  - https://www.medrxiv.org/content/10.64898/2026.08.27.26361548v1
AB  - BackgroundEmergency Department (ED) crowding is an international crisis primarily driven by exit block. Point of care (POC) cardiac biomarker testing and reduced sampling intervals have been proposed to mitigate crowding by improving throughput, but whole-ED operational impacts remain poorly understood, and evaluations often rely on vulnerable observational designs. This study aimed to assess whether introducing POC high-sensitivity troponin testing and reduced sampling intervals changed whole-ED flow metrics, and to test the robustness of interrupted time series (ITS) methodology in this setting.  MethodsA multi-centre controlled interrupted time series (CITS) across two large urban intervention EDs and one untreated control ED in Glasgow, UK. The intervention combined whole-blood POC high-sensitivity troponin testing with a reduction in sampling intervals from 3 to 2 hours. Outcomes included daily ED admissions, mean occupancy, maximum occupancy, and mean length of stay. Analyses used a window of 120 days either side of each implementation date. Effects were evaluated using segmented ITS models, with and without controls, with permutation tests against 147 pre-intervention placebo dates. The minimum detectable effects of a similar study, applied to a national dataset, were simulated.  ResultsAcross 483,412 presentations to the intervention sites, the intervention produced no statistically significant change in any whole-ED flow metric against the untreated control at either site. Analysed alone, one intervention site appeared to show reductions in mean occupancy (-6.08, 95% CI -12.04 to -0.12) and maximum occupancy (-7.60, -14.47 to -0.73); the untreated control department produced reductions in the same direction at the same date, and both estimates attenuated to the null once the control was applied. Under a pre-specified 14-day transition specification the reductions in the untreated department reached statistical significance while those at the treated site did not. The study was limited by power due to the study window and limited control pool. Simulation demonstrated that a national dataset has the potential to provide operationally feasible and clinically important findings.  ConclusionPOC cardiac biomarker testing and reduced sampling intervals did not detectably improve whole-ED flow, though the design was underpowered. More importantly, uncontrolled ITS designs are highly vulnerable to confounding in complex healthcare systems; evaluations of operational interventions must utilise concurrent controls, and routinely report falsification tests.  What is already known on this topicO_LIEmergency Department (ED) crowding is largely driven by exit block, yet throughput interventions like point of care (POC) testing are marketed to im- prove patient flow. C_LIO_LIReal-world effectiveness studies of rapid rule-out strategies for myocardial in- farction have not consistently demonstrated operational improvements, particu- larly in systems constrained by downstream pressures. C_LIO_LISingle, uncontrolled interrupted time series (ITS) designs are frequently used to evaluate complex healthcare interventions despite known vulnerabilities to un- measured confounding. C_LI  What this study addsO_LIImplementing POC high-sensitivity troponin alongside reduced sampling inter- vals did not detectably improve whole-ED admissions, occupancy, or length of stay, though the study was critically underpowered to detect plausible, smaller effect sizes due to the dilution of whole-ED metrics. C_LIO_LIUncontrolled analysis of one intervention site generated statistically significant apparent improvements in crowding that were not replicated in controlled analy- sis alongside an untreated department in the same health board, in which esti- mates of the same direction were obtained. C_LIO_LIConventional model-based tests declared an effect in large proportions of pre- intervention dates, so falsification testing rather than model-based inference alone is required to interpret these designs. C_LI  How this study might affect research, practice or policyO_LIMethodologically, researchers evaluating operational interventions in emer- gency care must be aware that uncontrolled ITS is vulnerable to confounding, consider designs using concurrent controls, and mandate the use of falsification (placebo) testing as routine practice. C_LIO_LIPolicymakers and clinicians should consider the potential for the influence of system-wide shocks when evaluating positive results from uncontrolled, single- site before-and-after or ITS studies. C_LIO_LIOperational managers should recognise that throughput interventions cannot overcome whole-ED crowding driven by exit block; resources must be directed accordingly. C_LI
ER  -