- Design
- multicentre, randomised, open-label, blinded-endpoint trial at 67 centres in China (HeadSOAR)
- Population
- 1,368 adults with anterior circulation large vessel occlusion stroke and successful reperfusion (eTICI 2b or better) after thrombectomy; median age 68, median NIHSS 15
- Primary outcome
- modified Rankin scale at 90 days, by shift analysis
- Effect
- adjusted generalised odds ratio 1.12 (95% CI 0.97-1.29; P = 0.14); 90-day mortality 18.3% vs 20.3% (adjusted RR 0.86; 95% CI 0.69-1.07)
HeadSOAR randomised 1,368 patients across 67 Chinese stroke centres, all of whom had achieved successful reperfusion after endovascular thrombectomy for anterior circulation large vessel occlusion, to 72 hours of head elevation at 30-40 degrees or flat positioning at 0-10 degrees. Median baseline NIHSS was 15, and 99.3% completed 90-day follow-up.
The shift analysis of 90-day modified Rankin scores gave an adjusted generalised odds ratio of 1.12 for lower disability with head elevation, which did not reach significance. Mortality at 90 days was 18.3% against 20.3%, also not significant.
The authors' own conclusion is unusually candid: the difference between groups was smaller than the trial was designed to detect, which leaves open a modest but clinically meaningful benefit that this trial could not resolve. That is the right way to read it. Both point estimates favour elevation, neither interval excludes no effect, and the intervention costs nothing.
For practice, this is permission rather than instruction. Nursing units that elevate the head after thrombectomy have no reason to stop; units that keep patients flat have no evidence they are harming anyone. What it does rule out is a large benefit in either direction, which is worth knowing before another trial is designed around the same question.
- Do not change an established unit protocol on the strength of this trial in either direction
- Where head position is being altered for another reason — aspiration risk, raised intracranial pressure — those reasons still govern
- Record the position actually maintained, since a protocol that is not followed produces exactly this kind of narrowed difference
- Read the confidence intervals rather than the p-values; both favour elevation without reaching significance
The statistics, in plain English
An adjusted generalised odds ratio of 1.12 with a 95% confidence interval of 0.97 to 1.29 means the data are compatible with anything from a very slight disadvantage to a 29% relative improvement in the odds of a better disability score. The interval only just crosses 1.0, which is why the authors describe a possible modest benefit rather than calling the intervention ineffective. A shift analysis uses the whole range of the modified Rankin scale rather than a single cut-point, which makes it more sensitive to small differences — and it still did not reach significance here.
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