← Back

AVERT

A Very Early Rehabilitation Trial after stroke (AVERT): a Phase III, multicentre, randomised controlled trial

Year of Publication: 2017

Authors: Langhorne P, Wu O, Rodgers H, ..., on behalf of the AVERT triallists' collaboration.

Journal: Health Technology Assessment (NIHR HTA monograph — full trial report; primary results also published in The Lancet 2015;386:46-55)

Citation: Langhorne P, Wu O, Rodgers H, Ashburn A, Bernhardt J. A Very Early Rehabilitation Trial after stroke (AVERT): a Phase III, multicentre, randomised controlled trial. Health Technol Assess 2017;21(54).

Link: https://doi.org/10.3310/hta21540

PDF: https://www.journalslibrary.nihr.ac.uk/hta/hta21540/

Clinical Question

Does very early mobilisation (VEM) — beginning within 24 hours of stroke onset at higher frequency and dose than usual care — improve functional independence at 3 months compared with standard stroke unit care?

Study Overview

Objective

To determine whether very early mobilization (within 24h) improves outcomes in acute stroke patients compared to standard care.

Study Summary

Early mobilization did not improve functional outcome at 3 months
Reduced functional independence (mRS 0-2) with early mobilization (46% vs 50%, aOR 0.73, p=0.004)
Early mobilization led to shorter hospital stays but more frequent and longer activity sessions

Intervention

Multicenter, open-label randomized trial comparing very early mobilization (median 18.5h after stroke) vs standard mobilization (median 22.4h). Intervention included more frequent and longer sessions of sitting, standing, and walking.

Patients per Arm

1054 early mobilization, 1050 usual care

Bottom Line

VEM within 24h was harmful: functional independence (mRS 0-2) at 3 months was significantly lower with VEM (46% vs 50%; adjusted OR 0.73; 95% CI 0.59-0.90; P=0.004). Dose-response analysis revealed a paradox: more frequent short sessions improved outcomes (OR 1.13/session), but more total daily minutes worsened outcomes (OR 0.94 per 5 min) — suggesting short, frequent mobilizations are preferable to prolonged early sessions.

        Major Points
        VEM significantly worsened outcome: 46% vs 50% mRS 0-2 at 3 months (aOR 0.73; 95% CI 0.59-0.90; P=0.004).
Intervention delivered: VEM patients mobilised 4.8h earlier, 3 more sessions/day, 21 more minutes/day (all P<0.0001). 92% vs 59% mobilised within 24h.
No significant mortality difference: 8% vs 7% at 3 months (OR 1.34; 95% CI 0.93-1.93; P=0.113).
No walking recovery difference: 75% vs 76% walking unassisted by 3 months (aOR 0.83, 95% CI 0.64-1.07, P=0.143); time-to-walk also similar (HR 1.04, P=0.459).
No QoL difference at 12 months: AQoL 0.47 vs 0.49 (P=0.865).
Dose-response paradox: frequency improved outcomes (OR 1.13/session; P<0.001) while duration worsened them (OR 0.94/5min; P<0.001).
ICH subgroup signal: OR 0.48 for favorable outcome, OR 3.21 for death (not significant interaction but concerning).
Prespecified severe-stroke subgroup (NIHSS >16, n=291): OR 0.35 (0.11-1.18) — direction favours UC but not statistically significant.
Shorter hospital stay with VEM: median 16 days vs 18 days total; acute-care-only stay 7 days in both groups.
UC temporal drift: TTFM shortened 28 min/year (P=0.001) over 8-year trial, narrowing group contrast.
Meta-analysis of 9 RCTs (2,618 patients): early mobilisation showed no benefit — OR 1.10 (0.94-1.29) for death/dependency.

      

Design

Study Type: Phase III pragmatic parallel-group randomized controlled trial

Randomization: 1

Blinding: Single-blind (blinded outcome assessors). Web-based stratified block randomization (average block size 6), stratified by site and NIHSS (mild 1-7, moderate 8-16, severe >16).

Enrollment Period: July 18, 2006 to October 16, 2014

Follow-up Duration: Primary: 3 months. Secondary: 12 months.

Centers: 56

Countries: UK, Australia, New Zealand, Malaysia, Singapore

Sample Size: 2104

Analysis: Intention-to-treat. Binary logistic regression adjusted for baseline NIHSS and age. Power: 80% to detect ≥7.1% absolute risk reduction.

Inclusion Criteria

Age ≥18 years with clinical diagnosis of first or recurrent stroke (infarct or ICH).
Admitted to hospital within 24 hours of stroke onset.
Admitted to acute stroke unit.
Able to react to verbal commands (minimum consciousness level).
Thrombolysis patients eligible if attending physician permitted mobilisation within 24h.

Exclusion Criteria

Prestroke mRS 3-5.
TIA.
Direct ICU admission, palliative decision, or need for immediate surgery.
Rapidly deteriorating disease (e.g., terminal cancer).
Lower limb fracture preventing mobilisation.
Concurrent recruitment to drug/intervention trials.
Unstable coronary/medical condition.
Unstable physiology: SBP <110 or >220, SpO2 <92%, HR <40 or >110, temp >38.5°C.

Baseline Characteristics

Characteristic	VEM (N=1,054)	Usual Care (N=1,050)
Age mean (IQR)	72.3 (62.3-80.3)	72.7 (63.4-80.4)
Female	411 (39%)	407 (39%)
Hypertension	707 (67%)	717 (68%)
Diabetes	239 (23%)	228 (21%)
AF	229 (22%)	237 (23%)
NIHSS median (IQR)	7 (4-12)	7 (4-12)
NIHSS mild (1-7)	592 (56%)	578 (55%)
NIHSS moderate (8-16)	315 (30%)	328 (31%)
NIHSS severe (>16)	147 (14%)	144 (14%)
ICH	142 (14%)	116 (11%)
IV tPA	247 (23%)	260 (25%)
Premorbid mRS 0	799 (76%)	786 (75%)
First stroke	878 (83%)	843 (80%)
Time to randomization mean (IQR)	18.2h (12.1-21.8)	18.2h (12.5-21.8)

Arms

Field	Very Early Mobilisation (VEM)	Control
Intervention	Out-of-bed sitting, standing, walking within 24h of stroke onset. At least 3 additional sessions/day beyond usual care. Titrated to 4 functional levels (MSAS score). Max seated time 50 min/session for first 3 days. 5-step safety protocol before first mobilisation. Continued 14 days or until stroke unit discharge. Actual: TTFM 18.5h, 6.5 sessions/day, 31 min/day OOB.	Standard stroke unit care per site discretion. Mobilisation not prescribed, only recorded. Actual: TTFM 22.4h, 3 sessions/day, 10 min/day OOB. 59% mobilised within 24h (vs 92% VEM).
Duration	14 days or until discharge	Standard care

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR	P-value
Favourable outcome (mRS 0-2) at 3 months	Primary	525/1,045 (50%)	480/1,038 (46%)		0.004
mRS ordinal shift at 3 months	Secondary	—	—	Adjusted generalised OR 0.94	0.193
Walking 50m unassisted at 3 months (proportion)	Secondary	796 (76%)	784 (75%)	aOR 0.83	0.143
Time to walk 50m unassisted (median days)	Secondary	7 (6-8); n=1049	6 (5-7); n=1051	aHR 1.04	0.459
Death at 3 months (case fatality)	Secondary	72 (7%)	88/1048 (8%)	OR 1.34	0.113
Non-fatal SAEs (incidence rate)	Secondary	≥1 in 20%	≥1 in 19%	IRR 0.88	0.194
AQoL at 12 months (median)	Secondary	0.49 (IQR 0.08-0.81)	0.47 (IQR 0.07-0.81)	Adjusted median difference -0.004	0.865
Length of hospital stay (median days, acute + rehab)	Secondary	18 (IQR 6-43)	16 (IQR 5-44)
Death at 3 months	Adverse	72 (7%)	88/1048 (8%)	OR 1.34 (0.93-1.93)	0.113
Non-fatal SAEs (≥1 event, all categories)	Adverse	208 (20%)	201 (19%)	IRR 0.88 (0.72-1.07)	0.194
Immobility-related SAEs (fatal + non-fatal; pulmonary embolism, DVT, UTI, pressure sores, pneumonia)	Adverse	53 (5%)	54 (5%)	IRR 0.92 (0.62-1.35)	0.665
Neurological SAEs (fatal + non-fatal; stroke progression, recurrent stroke)	Adverse	83 (8%)	107 (10%)	IRR 1.26 (0.95-1.66)	0.108
Stroke progression (total events)	Adverse	56 (5%)	72 (7%)
Pneumonia (fatal cases contributing to death)	Adverse	15	19
Recurrent stroke (fatal cases contributing to death)	Adverse	7	11

Subgroup Analysis

Prespecified subgroup analyses tended to favour UC across most main subgroups, though no interaction was statistically significant (all P>0.05). Key point estimates: mild NIHSS 1-7: OR 0.75 (0.57-0.98); moderate NIHSS 8-16: OR 0.76 (0.53-1.08); severe NIHSS >16 (prespecified, n=291): OR 0.35 (0.11-1.18); infarct: OR 0.77 (0.62-0.97); ICH: OR 0.48 (0.25-0.92) for favorable outcome and OR 3.21 (1.13-9.07) for death; rtPA-treated: OR 0.71 (0.46-1.09); TTFM <12h: OR 1.02 (0.62-1.68) — the only subgroup that was directionally neutral/slightly favouring VEM; TTFM 12-24h: OR 0.56 (0.42-0.75); TTFM >24h: OR 0.78 (0.42-1.43). (The '7% vs 35%' figure sometimes cited for severe stroke comes from an exploratory CART subset, not the prespecified severe-stroke subgroup.)

Criticisms

UC contamination: TTFM shortened 28 min/year (P=0.001) over 8-year trial; by end, ~2/3 UC patients mobilised within 24h.
Complex intervention difficult to standardize across 56 sites.
Unblinded treating staff could influence outcomes beyond mobilisation (attention, reassurance).
Physiological monitoring data limited in large pragmatic trial.
Subgroup interactions underpowered — ICH/severe stroke trends are hypothesis-generating.
Daily amount measured from physiotherapy only, may underestimate total OOB time.
Economic analysis not completed.

Funding

NIHR Health Technology Assessment programme (UK). Additional funding from National Health and Medical Research Council (NHMRC) Australia, Singapore Health, Chest Heart and Stroke Scotland, Northern Ireland Chest Heart and Stroke, and The Stroke Association (UK). NHMRC fellowship funding to Julie Bernhardt (1058635); Australian Research Council fellowship (0991086); National Heart Foundation Australia (G04M1571). The Florey Institute of Neuroscience and Mental Health (trial host) acknowledges support from the Victorian Government via the Operational Infrastructure Support Scheme.

Based on: AVERT (Health Technology Assessment (NIHR HTA monograph — full trial report; primary results also published in The Lancet 2015;386:46-55), 2017)