HeadPoST
Cluster-Randomized, Crossover Trial of Head Positioning in Acute Stroke
Bottom Line
Head positioning (lying-flat vs sitting-up ≥30°) initiated a median of 14 hours after acute stroke onset and maintained for 24 hours did not alter 90-day disability, mortality, or pneumonia risk.
Major Points
- Pragmatic international cluster-randomized crossover trial at 114 hospitals in 9 countries randomized 11,093 acute-stroke patients (5,295 lying-flat vs 5,798 sitting-up) to a head position initiated soon after admission and maintained for 24 hours.
- Primary outcome (90-day disability on modified Rankin scale) showed no difference: unadjusted OR 1.01 (95% CI 0.92–1.10), P=0.84; adjusted analyses and analyses restricted to ischemic stroke or ICH gave the same result.
- Death or major disability (mRS 3–6) at 90 days: 38.9% vs 39.7% (OR 0.94, 95% CI 0.85–1.05, P=0.25); 90-day mortality 7.3% vs 7.4% (OR 0.98, 95% CI 0.85–1.14, P=0.83).
- Serious adverse events (14.3% vs 13.5%, P=0.51), including pneumonia, did not differ; the sitting-up group achieved better position adherence (95% vs 87% maintained for 24 hours, P<0.001).
- Findings were consistent across prespecified subgroups (age, sex, region, baseline NIHSS, time from onset, stroke subtype, randomization sequence); most patients were enrolled well beyond the reperfusion window with mostly mild strokes (median NIHSS 4).
Design
Study Type: Pragmatic, cluster-randomized, crossover, open-label trial with blinded outcome assessment
Randomization: 1
Blinding: Open-label intervention; outcome assessors blinded (PROBE-style)
Enrollment Period: March 2, 2015 – November 29, 2016
Follow-up Duration: 90 days
Centers: 114
Countries: Australia, China, Taiwan, India, Sri Lanka, United Kingdom, Chile, Brazil, Colombia
Sample Size: 11093
Analysis: Intention-to-treat, ordinal logistic-regression mixed model with fixed intervention and period effects, random cluster effect, and cluster × period interaction; multiple imputation for missing mRS
Inclusion Criteria
- Age ≥18 years
- Clinical diagnosis of acute stroke (ischemic or intracerebral hemorrhage)
- Presentation to emergency department or inpatient service at a participating hospital
- Patients with acute intracerebral hemorrhage (but not subarachnoid hemorrhage) purposely included
- Ability to assume the assigned head position soon after admission
Exclusion Criteria
- Confirmed diagnosis of transient ischemic attack
- Clear clinical indication for, or contraindication to, either head position (e.g., patients likely requiring intubation)
- Local clinician-investigator judged the assigned head position could not be consistently maintained
- Subarachnoid hemorrhage
- Patient declined participation
Arms
| Field | Lying-flat | Control |
|---|---|---|
| Intervention | Fully supine (back horizontal, face upwards), initiated soon after admission and maintained for 24 hours, including during eating, drinking, and toileting | Head of bed elevated to ≥30° (confirmed by protractor), maintained for 24 hours; toilet privileges outside bed permitted per mobility |
| Duration | 24 hours | 24 hours |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Shift in distribution of 90-day disability on the modified Rankin scale (mRS 0–6), analyzed by ordinal logistic regression | Primary | Sitting-up mRS distribution (reference) | Lying-flat mRS distribution | 1.01 | 0.84 |
| Death or major disability (mRS 3–6) at 90 days | Secondary | 39.7% | 38.9% | 0.94 | 0.25 |
| All-cause death within 90 days | Secondary | 7.4% | 7.3% | 0.98 | 0.83 |
| EQ-5D visual-analogue self-rated health at 90 days | Secondary | reference | favored lying-flat | 0.009 | |
| Position adherence (maintained assigned position for 24 h) | Secondary | 95% | 87% | <0.001 | |
| Premature cessation of assigned position within 24 h | Secondary | 4.2% | 13.0% | <0.001 | |
| mRS distribution at 7 days / NIHSS categorical scores at 7 days | Secondary | reference | no significant difference | NS | |
| Serious adverse events (any) | Adverse | Lying-flat 14.3% vs Sitting-up 13.5% (P=0.51) | |||
| Pneumonia | Adverse | No significant between-group difference | |||
| Cardiopulmonary adverse events / oxygen saturation / blood pressure | Adverse | No significant between-group differences | |||
Subgroup Analysis
No significant heterogeneity in the primary outcome across prespecified subgroups defined by age, sex, country/region grouping, baseline NIHSS, time from stroke onset to intervention, major pathologic subtype (ischemic vs ICH), or hospital randomization sequence. Post-hoc analyses by NIHSS quintile and by time-to-intervention quintile also showed no heterogeneity.
Criticisms
- Median 14 hours from stroke onset to positioning meant most patients were past the reperfusion window and any effect of positioning on the ischemic penumbra would be minimal.
- Mostly mild strokes (median baseline NIHSS 4), limiting the ability to detect a benefit that might exist in more severe or large-vessel occlusion strokes.
- Cluster-randomized crossover design with unavoidable open-label intervention; adherence to the lying-flat position was significantly lower (87% vs 95%), potentially diluting a true effect.
- Target sample size (12,000) not reached; power reestimated using observed intracluster (0.083) and interperiod (0.076) correlations, but subgroup analyses remained underpowered.
- Included acute intracerebral hemorrhage in a stroke-positioning trial, which mixes pathophysiologies; separate analyses by subtype were nonetheless consistent with the primary result.
- Enrollment concentrated in centers largely outside the acute reperfusion era limits generalizability to modern thrombolysis/thrombectomy workflows.
Funding
National Health and Medical Research Council of Australia (research grant 1066966); coordinated by the George Institute for Global Health.
Based on: HeadPoST (NEJM, 2017)
Authors: Anderson CS, Arima H, Lavados P, et al. (HeadPoST Investigators and Coordinators)
Citation: N Engl J Med 2017;376:2437–2447. DOI: 10.1056/NEJMoa1615715
Content summarized and formatted by NeuroTrials.ai.