INFORMATION FROM HEADPOST AND AVERT (positioning and mobilisation strategies within 24 hrs)

PETER LANGHORNE UNIVERSITY OF GLASGOW

Sincere thanks to:

Caroline Watkins and HeadPoST investigators Julie Bernhardt and AVERT investigators

EARLY POSITIONING AND MOBILISATION

Head Position in Stroke Trial (HeadPoST) results: implications for clinical practice

Dr Liz Lightbody On behalf of the Steering, Operations and Advisory Committees for Investigators and Coordinators across 114 hospitals in 9 countries

An international collaborative project

Main funding support

Other funding support

An international collaborative project

Potential benefits

• Reduced risk of aspiration pneumonia

• Avoid delayed mobilisation

• Better cardiorespiratory dysfunction

• Reduce Intracranial Pressure (ICP) and oedema

Potential harms

• Impairment in the cerebral autoregulation

• Progressive neurological deterioration following admission

Sitting up in first 24 hours

Meta-analysis of mean cerebral blood flow velocity by TCD in different head positions in pts with acute ischemic stroke

Figure 1. Effects of head position change on mean flow velocity

Overall estimates of effect and 95%CI were calculated using a fixed effects model and inverse variance weighting. Boxes

and horizontal lines represent mean difference and 95% CI for each trial; size of boxes is proportional to the inverse of

variance of the trial results. Diamonds show 95% CI for pooled estimates of effect and are cantered on pooled mean

difference. HOBOE indicates Head-of-Bed Optimization of Elevation Study

Mean difference (95% CI)

Mean difference (cm/s)

IncreaseDecrease

-10 0 20 40

6.00 (-19.87 to 31.87)

5.00 ( -8.92 to 18.92)

4.80 ( 2.84 to 6.76)

1.10 ( -4.67 to 7.18)

4.47 ( 2.62 to 6.31)

11.60 (-16.62 to 39.82)

7.00 ( -7.70 to 21.70)

8.20 ( 4.28 to 12.12)

11.30 ( -1.64 to 24.24)

8.42 ( 4.82 to 12.03)

From 30 to 15

Schwarz (2002)

Wojner (2002)

Wojner-Alexandrov (2005)

HOBOE study (2011)

OVERALL

Homogeneity: chi-squared=1.31, df=3 (p=0.73), I2=0%

Test for overall effect: Z=4.75 (p<0.001)

From 30 to 0

Schwarz (2002)

Wojner (2002)

Wojner-Alexandrov (2005)

HOBOE study (2011)

OVERALL

Homogeneity: chi-squared=0.29, df=3 (p=0.96), I2=0%

Test for overall effect: Z=4.58 (p<0.001)

Olavarrria V, Arima H, Brunser, A, Anderson C, Lavados P ()

To determine benefits and risks of lying-flat vs.

sitting-up (≥30 degrees) head positions

- applied early and continued for 24 hours

- broad range of hospitalized stroke patients

HeadPoST aims

Design - sample size and statistics

70 patients per cluster (n=140) across 140 centers - 16,800 pts (14,000 AIS + 2,800 ICH)

90% power ( 0.05) - detect ≥16% (shift) effect in outcome, ordinal logistic regression

Conservative assumptions: 10% drop-out, 5% crossover, 10% lost-to-follow-up of patients, at each center; ICC (relation of clusters) 0.03, no IPC (relation of treatment periods per cluster)

Blinded outcome assessment at 90 days

Lying-flat (0°) Sitting-up (≥30°) Standard

nursing

+

medical

care Crossover Crossover

Randomization

Lying-flat (0°) Sitting-up (≥30°)

Hospital centers

Chile (7 sites, 608 pts)

China (39 sites; 4479 pts)

Brazil (4 sites, 264 pts)

UK (41 sites, 4160 pts)

Colombia (1 site, 38 pts)

Taiwan (5 sites; 173 pts)

HeadPoST network – 11,093 patients recruited

from 114 centers between March 2015 and August 2016

India (6 sites; 499 pts)

Sri Lanka (4 sites; 271 pts)

Australia (7 sites, 602 pts)

Baseline - Demographic and clinical

Variable

Lying-flat

N=5295

Sitting-up

N=5798

Age, mean 68 68

Female 40% 40%

Aust/UK 42% 44%

Hypertension 51% 51%

Prior stroke 23% 24%

Diabetes 20% 20%

Blood pressure, mean 155/87 155/87

NIHSS median (iqr) score 4 (2-9) 4 (2-9)

Time from stroke onset, median (iqr), hr 14 (5–35) 14 (5-25)

8

Thrombolysis, rtPA 12% 12%

Adherence to intervention

Lying-flat

(N=5295)

Sitting-up

(N=5798)

Odds ratio/Mean

difference (95% CI)

P value

Time spent in allocated position - hr

Median (iqr) 23 (20-24) 24 (23-24) -1.6 (-2.1 to -1.2) <0.0001

Intervention discontinued - n (%) 695 (13) 245 (4) 4.0 (3.1 to 5.3) <0.0001

Not tolerated 201 (29) 21 (9)

Unable to comply 73 (11) 11 (5)

Patient preference 135 (19) 37 (15)

Doctor preference 35 (5) 5 (2)

Change in medical condition 85 (12) 11 (5)

Other 166 (24) 158 (65)

Not specified - 2 (1)

Primary clinical outcome Death or disability (mRS 2-6) at 90 days

15.9

18.2

36.4

33.6

8.8

8.6

15.2

16.2

9.5

8.8

6.1

6.4

8.1

8.2

0 10 20 30 40 50 60 70 80 90 100

LYING FLAT

SITTING UP

Percentage of mRS scores

0 1 2 3 4 5 6

Unadjusted Odds ratio 1.01 (95%CI 0.92 to 1.10) P=0.84

Adjusted baseline demographic variables Odds ratio 1.05 (95%CI 0.96 to 1.15) P=0.30

Adjusted baseline demographic + risk factors Odds ratio 1.03 (95%CI 0.94 to 1.13) P=0.55

Adjusted with imputation Odds ratio 1.03 (95%CI 0.94 to 1.13) P=0.50

Not significant for acute ischemic stroke or intracerebral hemorrhage

HeadPoST Subgroup of primary outcome Shift-mRS at 90 days

In acute stroke patients, positioning either lying-flat or sitting-up in the first 24 hours of hospitalization:

• Lying flat is uncomfortable

• No differences in effects on disability outcome

• Positioning is safe - no differential SAEs or pneumonia

• No clear benefits (or harms) in subgroups according to: time from onset of symptoms initial stroke severity age region AIS vs. ICH / subtype of ischemic stroke

Major findings of HeadPoST

Successful completion of a novel ‘cluster crossover’ design to evaluate a nursing management strategy

No clear benefits (or harms) of specific head positioning in acute stroke

Anderson et al. NEJM 2017;376:2437-47

Implications for Practice

Early mobilisation: when, who and how? - AVERT

Peter Langhorne,

Professor of stroke care,

Glasgow University

What is Very Early Mobilisation (VEM)?

• Protocol developed by the AVERT research team.

• Based on treatment provided in stroke unit in Trondheim, Norway

• Interventions provided by a nurse /physiotherapist team

• Protocol/training provided to staff

• Tested in pilot trials

2104 participants recruited

56 centres (8 countries)

2006 - 2014

Complete follow up 2083 (99%)

Intervention achieved significant differences VEM Usual Care

median, IQR n=1054 n=1050 p value median shift (95% CI)

Time to first mobilisation (hrs) 18·5 (12·8–22·3) 22·4 (16·5–29·3) <0·0001 4·8 (4·1–5·7)

n=1042; missing=12 n=1036; missing=14

Frequency per person

(median daily sessions

of out of bed activity)

6·5 (4·0–9·5) 3 (2·0–4·5) <0·0001 3 (3–3·5)

Daily amount per person*

(median minutes per day

spent in out of bed activity)

31 (16·5–50·5) 10 (0–18) <0·0001 21·0 (20–22·5)

Total amount per person

(mins over the intervention period) 201·5 (108–340) 70 (32–130) <0·0001 117 (107–128)

75% of all patients started out of bed activity <24 hours

Intervention achieved significant differences VEM Usual Care

median, IQR n=1054 n=1050 p value median shift (95% CI)

Time to first mobilisation (hrs) 18·5 (12·8–22·3) 22·4 (16·5–29·3) <0·0001 4·8 (4·1–5·7)

n=1042; missing=12 n=1036; missing=14

Frequency per person

(median daily sessions

of out of bed activity)

6·5 (4·0–9·5) 3 (2·0–4·5) <0·0001 3 (3–3·5)

Daily amount per person*

(median minutes per day

spent in out of bed activity)

31 (16·5–50·5) 10 (0–18) <0·0001 21·0 (20–22·5)

Total amount per person

(mins over the intervention period) 201·5 (108–340) 70 (32–130) <0·0001 117 (107–128)

9.2

8.7

19.5

19.3

21.5

18.2

21

23

12

14

10

9

7

8

Usual care

VEM

mRS=0 mRS=1 mRS=2 mRS=3 mRS=4 mRS=5 mRS=6

46%

50%

OR 0.73, 95% CI 0.59–0.90, P=0.004

VEM

Usual care

Independent

Primary outcome - independence (mRS 0-2)

Early Mobilisation trials Death or poor outcome (dependent) at 3 months

EM improves outcome

1.07 (0.91-1.25)

Cochrane review (update in progress)

Who – who is at risk of hazard?

Early SAEs in EM trials

Early deaths AVERT (n=2104)

Other trials (n=459)

Proportion died by 14 days 4.5% vs 3% 6% vs 3%

Odds ratios for early death

Total 1.8 (1.1–2.9) 2.2 (0.9-5.4)

Age >80yrs 2.4 (1.1–5.1) 5.0 (1.0-26)

Intracerebral haemorrhage 4.2 (1.1–16) 7.8 (0.3-263) * * Very few events

Stroke-related cause 29 vs 16 1.8 (0.8-3.9) * * Included non-fatal events

Bernhardt et al. Cerebrovasc Dis 2016;42(suppl 1): P36

Conclusions (1)

1. Very early, higher dose out of bed activity protocol reduced the odds of favourable outcome, without accelerating walking recovery or reducing immobility-related SAEs

2. Early hazard confirmed (older age, intracerebral haemorrhage, stroke-related causes?)

3. Better outcomes seen with short, frequent mobilisations in the early phase after stroke (more likely to regain independence)

4. We need to characterise the “optimal” patterns of mobilisation for these patients

Early hazard may be due to neurological complications in more frail individuals?

When and how - so how do we unpack this?

Bernhardt IJS 2015 SAP

Regression models: eg. Efficacy 3 dose characteristics as independent variables, 3 months mRS outcome (dichotomized mRS 0-2 favourable outcome vs mRS 3-6 as poor outcome) as the dependent variable, adjusted for baseline NIHSS and age Classification and regression tree CART®analyses

Prespecified dose response analysis – all patients, no group

Key intervention criteria: 1) Time to first mobilisation (hours after stroke onset) 2) Frequency, median sessions per day (nurse, PT data) 3) Amount activity out of bed, median minutes per day (PT only)

When and how - so how do we unpack this?

Bernhardt IJS 2015 SAP

Regression models: eg. Efficacy 3 dose characteristics as independent variables, 3 months mRS outcome (dichotomized mRS 0-2 favourable outcome vs mRS 3-6 as poor outcome) as the dependent variable, adjusted for baseline NIHSS and age Classification and regression tree CART®analyses

Prespecified dose response analysis – all patients, no group

Key intervention criteria: 1) Time to first mobilisation (hours after stroke onset) 2) Frequency, median sessions per day (nurse, PT data) 3) Amount activity out of bed, median minutes per day (PT only)

Better outcomes seen with short, frequent mobilisations in the early phase after stroke

(more likely to survive and regain independence and walking)

Conclusions (2)

Further exploration will help underpin clinical practice guidelines

1. Very early, higher dose out of bed activity protocol reduced the odds of favourable outcome, without accelerating walking recovery or reducing immobility-related SAEs

2. Early hazard confirmed (older age, intracerebral haemorrhage, stroke-related causes?)

3. Better outcomes seen with short, frequent mobilisations in the early phase after stroke (more likely to regain independence)

4. We need to characterise the “optimal” patterns of mobilisation for these patients

Watch this space