- Design
- open-label, pragmatic, randomised controlled trial with economic evaluation (RESHAP-ED)
- Population
- 1491 adults with uncomplicated musculoskeletal conditions at five emergency departments in New South Wales, Australia; mean age 40.1 years
- Primary outcome
- emergency department length of stay, arrival to discharge
- Effect
- 2.4 h vs 3.4 h, difference -1.0 h (95% CI -1.2 to -0.8); adverse events 6.6% vs 5.9% (p=0.72); costs lower by A$35.1
RESHAP-ED randomised 1491 adults presenting to five emergency departments in New South Wales with uncomplicated musculoskeletal problems - soft tissue injuries, neck and back pain, and fractures or dislocations not needing orthopaedic review or surgery - to physiotherapist-led care or usual physician or nurse practitioner care. Mean age was 40.1 years and follow-up for the primary outcome was complete for every participant.
Mean length of stay was 2.4 h with physiotherapist-led care against 3.4 h with usual care, a difference of 1.0 h (95% CI 0.8 to 1.2). Adverse events were reported by 6.6% and 5.9% respectively (p=0.72). Costs were lower by about A$35 per patient, with a 98.1% probability of being cost-effective on the trial's own economic analysis.
The trial is open-label, and length of stay is a process measure rather than a patient outcome - but it is the process measure that most departments are actually failing on, and the safety data are reassuring rather than merely absent. The catch is eligibility: 2728 of 4219 screened patients were ineligible, so this model handles a well-defined slice of the department's work rather than its general load. Where it is worth thinking about is exactly the setting with long waits and a heavy musculoskeletal case mix, which describes a great many Indian emergency departments - though it needs a physiotherapy establishment that is rostered to the department, not borrowed from the wards.
- Define the eligible presentations tightly in advance - this worked because the case mix was screened, not because physiotherapists saw everyone.
- Exclude anything needing orthopaedic review or surgery at triage, as the trial did.
- If you pilot this, measure arrival-to-discharge time and adverse events, the two things the trial measured.
- Rostering matters more than training: the model depends on a physiotherapist present in the department, not on call.
- Do not extend the result to complicated musculoskeletal presentations; nearly two-thirds of screened patients were ineligible.
The statistics, in plain English
A one-hour reduction with a confidence interval of 0.8 to 1.2 h is precise, and complete follow-up on the primary outcome means it is not an artefact of missing data. The safety comparison is the weaker part: 39 versus 33 events with p=0.72 is a null result on small numbers, which does not exclude a real difference in either direction. Cost-effectiveness stated as a 98.1% probability comes from a modelled distribution rather than a measured frequency, and it reflects the salary and cost structure of the Australian health system, not any other.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for top clinical updates, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free