Systematic review When should physical rehabilitation commence after stroke: a systematic review Elizabeth Lynch1*, Susan Hillier1, and Dominique Cadilhac2,3 Background Knowing when to commence physical rehabilitation after stroke is important to ensure optimal benefit for stroke survivors and efficient health care. The aims of this review were to: determine the effects on mortality, function and complications when physical rehabilitation commences ‘early’ (within seven days of stroke); and describe the effects of early transfer to rehabilitation wards/hospitals when sustained rehabilitation is unavailable in acute stroke units. Review summary From 3751 potential articles we included 5 randomized controlled trials and 38 cohort studies. Metaanalysis was performed with 3 randomized controlled trials involving 159 people to investigate the effects of commencing physical rehabilitation within 24 h of stroke compared to 48 h. Commencing physical rehabilitation within 24 h trended towards greater mortality (Mantel-Haenszel odds ratio 2·58; 95% confidence interval 0·98 to 6·79, P = 0·06), with no differences in complications or health outcomes. The cohort studies provided evidence of benefits when physical rehabilitation was commenced on the day of admission (n = 1), within 3 days of stroke (n = 3), or ‘sooner rather than later’ (3 of 4 studies). The effect of earlier transfer to rehabilitation was reported in 32 cohort studies. In 23/26 (88%) cohort studies that accounted for age and stroke severity, results favored earlier transfer for improving post-stroke function, with no consensus on timeframes. Conclusion In summary, the benefits of commencing physical rehabilitation within 24 h of stroke remain unclear from the current literature. Commencing physical rehabilitation or transferring to rehabilitation services ‘early’ may provide better functional outcomes. Key words: early ambulation, outcome assessment, rehabilitation, stroke, therapy, time factors
Introduction Rehabilitation after stroke consists of individually tailored interventions which reduce stroke-related activity limitations and participation restrictions (1). An important issue being debated is when physical rehabilitation (defined as physical therapy, Correspondence: Elizabeth Lynch*, International Centre for Allied Health Evidence, School of Health Sciences, Sansom Institute for Health Research, University of South Australia, GPO Box 2471, Adelaide, SA 5001, Australia. E-mail: [email protected] 1 International Centre for Allied Health Evidence, School of Health Sciences, Sansom Institute for Health Research, University of South Australia, Adelaide, SA, Australia 2 Stroke Division, The Florey Institute of Neuroscience and Mental Health, Heidelberg, Vic., Australia 3 Stroke and Ageing Research Centre, Department of Medicine, Monash University, Clayton, Vic., Australia Received: 31 October 2013; Accepted: 16 December 2013 Conflict of interest: None declared. DOI: 10.1111/ijs.12262 © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
occupational therapy and physical exercises aimed at preserving or restoring physical function) (2) should commence after stroke. It has been postulated that physical rehabilitation commenced too early may increase the area of brain ischaemia and result in worse functional outcomes (3). Conversely, it has also been hypothesized that early physical rehabilitation may promote positive brain function reorganization and recovery (4), so a delay in rehabilitation may result in less effectiveness. In addition, it is well established that people with stroke are extremely inactive in hospital (5) and this increases the risk of losing muscle mass, strength and function (6). Therefore, a delay in commencing physical rehabilitation may result in the person with stroke developing more secondary impairments. Best practice guidelines are unanimous in recommending stroke unit (SU) care for people with stroke, with early rehabilitation being a defining component of a SU. However, the definition of ‘early’ rehabilitation is somewhat ambiguous. The Australian clinical guidelines for stroke recommend that rehabilitation should commence the first day after stroke (7), whereas European guidelines simply recommend rehabilitation should commence ‘early’ (8) and the United States guidelines recommend that rehabilitation should commence as early as possible, once medical stability has been attained (9). Mobilization is an important component of physical rehabilitation, and guidelines from the United Kingdom specify that mobilization should be commenced within 24 h of stroke (10). The recommendations in these international stroke guidelines have been based on one randomized controlled trial (RCT) (11), which was also the only study included in the 2009 Cochrane review (12). Different models of SUs exist whereby some are comprehensive stroke units, which are staffed to provide acute care as well as rehabilitation over a period of weeks (13,14) while others are acute stroke units, where patients are admitted acutely and discharged within approximately 3 to 10 days (13,14). Therefore, people admitted to acute SUs who require rehabilitation beyond this time point need to be transferred to a separate rehabilitation facility if inpatient rehabilitation is required. Knowing how soon a person with stroke should be transferred to a rehabilitation hospital if an acute hospital is unable to provide this therapy is important. The aims of this systematic review were to determine when it is best to commence physical rehabilitation post-stroke. Specifically: 1. To update: What are the effects on mortality, function and complications when physical rehabilitation is commenced ‘early’ i.e. within seven days of stroke? 2. When sustained rehabilitation is not provided by an acute SU, what are the effects of earlier transfer to a designated rehabilitation ward/hospital? Outcomes of interest in this review were differences in the health and functional outcomes of people with stroke. Vol ••, •• 2014, ••–••
1
Systematic review
E. Lynch et al.
Methods
Results
Data sources A computerized search was initially conducted by a single reviewer (EL) from inception to December 2012: databases Scopus, CINAHL, Medline, Embase, The Cochrane Library, and Pubmed were searched. The search strategy used a combination of the free text terms ‘stroke’ (sub-headings rehabilitation and therapy), ‘Cerebrovascular accident’ (sub-headings rehabilitation and therapy), ‘outcome assessment’, ‘time factors’, ‘outcomes health care’, ‘rehabilitation’ ‘commencement of rehabilitation’, ‘early ambulation’ and ‘early rehabilitation’.
The search yielded 3751 titles and abstracts; of which 48 articles were retained for the review (Fig. 1).
Study selection Any RCT, systematic reviews, prospective or retrospective controlled clinical trials or cohort studies published in English meeting the following criteria were included: • Examined people with stroke who received inpatient physical rehabilitation • Compared groups of people with stroke who commenced physical rehabilitation at different time intervals following stroke either in SUs or via transfer to rehabilitation units • Collected any objective outcome measures, for example mortality, complications, Functional Independence Measure (FIM), Barthel Index (BI), length of stay, discharge destination. Data collection and analysis One reviewer (EL) examined titles and abstracts against the inclusion criteria. Following removal of duplicates, full text articles were obtained for a more detailed review by two independent reviewers (EL and SH) to assess for inclusion. Reference lists were scrutinized for further relevant articles. In cases where further details were required, one reviewer (EL) contacted the articles’ authors. Data were extracted from papers that met the review criteria, and the risk of bias of each study was assessed using the Critical Appraisal Skills Program (CASP) criteria for RCTs and cohort study designs (15). One reviewer (EL) assessed each paper for risk of bias, a second reviewer (SH) assessed all RCTs and randomly assessed one third of the cohort studies to check for consistency. Where there was sufficient homogeneity of study design (RCT), population, intervention and measures, a meta-analysis was performed using the review software package Revman5 (16). This required the identification of the number of participants in each group for each trial, as well as the total number (for dichotomous data) and the mean and standard deviations for each group (for ordinal scale data). Data were analysed as odds ratios or mean differences using fixed effects with 95% confidence intervals. Where the means and standard deviations required to perform a meta-analysis were not published, these were requested from the authors. Only data from people who survived their stroke were included in the meta-analysis of functional independence, as mortality outcomes were analysed separately. Where metaanalysis was not possible, the results from the included papers were synthesized narratively.
2
Vol ••, •• 2014, ••–••
What are the effects of commencing physical rehabilitation within 7 days of stroke? Sixteen papers were included to answer this question (11,22–36) (Table 1). Six of the RCT papers had results from the same study cohort (11,22–26) so were analysed as one study. The included RCTs were generally of good quality with a low risk of bias (Table 2), however were based on small samples sizes and may have been underpowered to answer particular research questions related to outcomes. Commencing physical rehabilitation within 24 h Three RCTs used similar protocols to commence physical rehabilitation early on SUs, and compared this to usual SU care in Australia, United Kingdom and Norway (11,29,30). Overall, 159 people with stroke were included in early (versus usual SU care) rehabilitation trials. The main aim of these three studies was to assess the benefits of mobilizing patients within 24 h of stroke onset (11,29) or hospital admission (30) compared with usual care which was within 48 h of stroke. In two of the studies, the frequency of mobilization in the early rehabilitation group was increased (11,29). All studies provided outcomes at 3 months for case fatality, complications, functional independence using the BI and ‘good outcome’ indicated by a modified Rankin Score (mRS) of 0–2 (equivalent to none or slight disability). The comparator groups at baseline were mostly similar in terms of factors affecting post-stroke recovery such as age, comorbidities and stroke severity, and casemix adjustment was not conducted in any of these studies. We were able to conduct meta-analyses using these data and results of our meta-analyses are presented in Figs 2–5. We found a trend for fewer deaths in the usual care group which was mobilized 24–48 h post stroke [P = 0·06, OR 2·58 (95% CI 0·98– 6·79), Fig. 2]. At 3 months there was no significant difference between groups in BI [P = 0·23, OR 1·20 (95% CI – 0·77–3·18)], Figure 3. The likelihood of having a good outcome and experiencing no complications within 3 months following stroke was similar between groups, (P = 0·66 and P = 0·82 respectively, Figs 4 and 5). The mobilization sessions appeared to be well tolerated, since in the one study reporting blood pressure changes during mobilizations, no patient had 3 consecutive drops in blood pressure >30 mm Hg during the first 3 attempted mobilizations (11). Commencing physical rehabilitation within first 7 days of stroke (Table 1) One RCT provided evidence that commencing physical rehabilitation 3 days compared to 7 days after stroke resulted in significantly fewer serious complications, and did not influence cerebral blood flow (27). Authors of a previously conducted RCT which compared SU care to general ward care (37) conducted a follow-up analysis and reported that after adjusting for age and severity of stroke, the time at which physical rehabilitation commenced (on day of admission on SU vs day 2 or 3 on medical ward) was an independent factor significantly associated with discharge home within 6 weeks of stroke (28). © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Systematic review
E. Lynch et al.
Records iden fied through database searching (n = 3751)
Records a er duplicates removed (n = 3127)
Records screened (n = 3127)
Addi onal relevant studies iden fied through hand searching of literature (n = 12)
Full-text ar cles assessed for eligibility (n = 87)
Ar cles included in qualita ve synthesis (n = 48, 43studies) Prospec ve observa onal cohort studies (n = 15) Retrospec ve observa onal cohort studies (n = 23) Randomised controlled trials (n = 10, 5 studies)
Records excluded (n = 3052)
Full-text ar cles excluded = 32 1. Study design and repor ng • No comparison of groups (n = 4) • No comparison/record regarding stroke onset-rehabilita on commencement interval (n = 6) • Comparing different models of care, effect of ming not analyzed (n = 7) • Review ar cle where ming of rehabilita on not main focus (n = 9) • Func onal outcomes not collected (n = 1) 2. Popula on • Not first rehabilita on admission for current stroke (n = 1) • Not inpa ent physical rehabilita on (n = 4) + Eligible ar cles not included = 7 1. Presented same data as another ar cle included in review (14, 17–21) (n = 6) 2. Review ar cle without systema c search (3) (n = 1)
Studies included in quan ta ve synthesis (meta-analysis) (n = 3)
Fig. 1 Process for identification of the included studies.
Six cohort studies (31–36) variously described the effect of timing of physical rehabilitation commencement on patient outcomes. Three of 4 cohort studies which used regression analyses to evaluate the effect of timing of rehabilitation commencement on BI, provided evidence that a better BI score on discharge was significantly associated with the timing of physical rehabilitation initiation (smaller delay, greater BI) (31–34). Authors of two cohort studies analysed the effects of commencing physical rehabilitation within 3 days (compared to 4 or more days), and reported that commencing within 3 days had no significant association with mortality (36), improved the likelihood of reducing disability by 15·3% (P < 0·001) (36); and resulted in better ambulation and ADL function, rated via non-validated outcome measurement tools (35). What are the effects of earlier transfer to the rehabilitation service? Thirteen prospective (38–50) and 19 retrospective observational cohort studies (51–69) met the inclusion criteria for this question. © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
In general, the quality of the studies regarding the effects of earlier transfer to a rehabilitation service after stroke was poor with a high risk of bias identified in a number of the included studies (Table 3). Studies were highly heterogeneous, with authors using different reporting methods, outcome measures and analyses (Table 4). The average time at which people with stroke were transferred to rehabilitation ranged from 11 days (39) to 75 days post-stroke (60). All 32 studies had some measure of clinical outcome, the most common being FIM (16 studies) or BI (7 studies). Twentysix of the 32 included studies (81%) accounted for age and stroke severity in the calculations relating functional outcome with the timing of transfer to rehabilitation. Among these 26 studies, the majority (n = 23, 88%) had results that favored early transfer to rehabilitation with regards to better functional outcome. The optimal time to transfer to the rehabilitation service was calculated in one study (61) and transfer within 12 days of stroke was recommended for best outcome based on a trend analysis of LOS efficiency (change in FIM divided by LOS). Vol ••, •• 2014, ••–••
3
4
Vol ••, •• 2014, ••–••
2010
2012
2011
1999 1991
2003
2010
2009
1986
2011
2010
Langhorne et al. (29)
Sundseth et al. (30)
Diserens et al. (27)
Indredavik et al. (28) Indredavik et al. (37)
Horner et al. (32)
Hu et al. (31)
Huang et al. (34)
Hayes & Carroll (35)
Kuwabara et al. (33)
Matsui et al. (36)
Japan
Japan
USA
Taiwan
Taiwan
USA
Norway
Netherlands
Norway
5482
45014
30
76
154
738
206
42
56
32
71
Australia
2008 2009 2008 2010 2010 2008
Bernhardt et al. (11) Sorbello et al. (24) Cumming et al. (22) Cumming et al. (23) Tyedin et al. (26) Tay-Teo et al. (25)
United Kingdom
n=
Country
Date
Author
Retrospective cohort study Retrospective cohort study
Post-hoc analysis from single centre RCT Prospective cohort study Prospective cohort study Retrospective cohort study Retrospective cohort study
Single centre RCT
Single centre RCT
Single centre RCT
Multi-centre RCT, Phase II
Study type
In-hospital mortality mRS ≤ 1 at discharge
3 days (delayed)
Not presented
LOS ADL Gait ability BI
BI at 3 months
BI at discharge
No complications in hospital mRS (0–2) PSV at 72–96 h EDV at 72–96 h Factors associated with discharge home within 6 weeks of stroke BI at 1 year
Mixed model multiple regression: Estimate −1·79, SE 0·55, P < 0·001 Multiple linear regression: β = −0·65, 95% CI −1·2 to −0·10, adjusted R2 = 0·64, P = 0·02 Multivariate analysis by stepwise linear regression: β = −2·45, SE = 0·5, P < 0·01 early 24 ± 12 days vs delayed 39 ± 15, P* = 0·005 early 2·40 ± 0·83 vs delayed 2·13 ± 0·74 early 2·87 ± 1·30 vs delayed 2·73 ± 1·30 Not significant: One day delay in initiating rehabilitation: B = 0·003, 95% CI −0·042 to 0·047 Early 1·7% vs delayed 1·6%, univariate P = 0·799 Favors commencing early: Marginal effect 0·048, 95% CI 0·013 to 0·084.
EM 7/27 (25·9%) vs SC 2/29 (6·9%), OR 4·73; 95% CI, 0·89–25·21; P = 0·07 EM 10/25 (40%) vs SC 17/28 (60·7%), OR 2·32, 95% CI, 0·77–6·98; P = 0·14 EM 9/27 (33·3%) vs SC 10/29 (24·4%); P = 0·93 EM median 19, IQR 7–20, SC median 19, IQR 6–20 Mean (SD) EM 3·9 (3·8), SC 5·5 (2·9), P = 0·02 3 days 18/25 (72%) vs 7 days 9/17 (53%) P* = 0·33 3 days 10/25 (40%) vs 7 days 6/17 (35%) P* = 1·00 3 days 75 ± 6 vs 7 days 73 ± 6 3 days 30 ± 3 vs 7 days 30 ± 3·5 Time to start of mobilization, coefficient −0·4862, relative risk 0·62, CI 0·45–0·84
Death mRS (0–2) No complications BI Improvement in NIHSS
Death mRS (0–2) No complications BI
EM 8/38 vs SC 3/33; absolute risk difference = 12%; 95% CI −4·3% to 28·2%, P = 0·20 EM 15/38 (39·5%) vs SC 10/33 (30·3%), P 0·46. EM 5/38 vs SC 8/33, P* = 0·36 EM median 16·5, IQR 2·0 to 20·0, SC median 16·5, IQR 9·0 to 20·0, P = 0·713 EM 10 (5–17) vs SC 12 (4–15), z = −0·057, P = 0·755 EM median 0·32 vs SC 0·24, P = 0·17 EM AUD 13 559 ± 12 667 vs SC AUD 21 860 ± 16 272, P = 0·02 EM 0/16 vs SC 1/16 EM 12/16 (75%) vs SC 7/16 (43·8%), P 0·07 EM 7/16 (43·8%) vs SC 4/16 (25%), P* = 0·46 EM median 20, IQR 18–20, SC median 17, IQR 2–20, P = 0·21
results
Death mRS (0–2) No complications BI IDA scale AQoL at 12 months Total cost
outcomes (at 3 months unless otherwise stated)
3 days (delayed)
Mean 6·7 ± 6·7 days (range 1–50 days) Mean 7·7 ± 4·6 days
Day of admission (SU) vs day 2 or 3 (general ward) Median 3 days
At 3 days (early) vs 7 days (delayed)
Within 24 h of stroke vs 24–48 h. Also increased frequency of mobilization in EM group Within 24 h of hospital admission vs 24–48 h
Within 24 h of stroke vs 24–48 h. Also increased frequency of mobilization in EM group
Time rehabilitation commenced
Table 1 Summary date from studies investigating Question 1: What are the effects of commencing physical rehabilitation within 7 days of stroke?
Systematic review E. Lynch et al.
© 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Systematic review
✓ ✓ ✓ x ✓ ✓ ✓ ✓ ✓ x ✓ ✓ ✓ ✓ unclear ✓ ✓ ✓ ✓ ✓ Bernhardt 2008 (11) Langhorne 2010 (29) Indredavik 1999 (28) Sundseth 2012 (30) Diserens 2011 (27)
✓ ✓ y ✓ ✓
✓ ✓ ✓ ✓ ✓
✓ ✓ s ✓ ✓
✓ ✓ ✓ x ✓
✓ ✓ ✓ ✓ ✓
Small Small Not applicable nil small
✓ s ✓ ✓ ✓
Benefit worth harms & costs Precision of results Treatment effect size Same management (aside from intervention) Similar at baseline Assessors blinded Patients accounted for Random assignment Focussed issue
Table 2 Critical appraisals of included randomized controlled trials using Critical Appraisal Skills Programme criteria (15)
Locally applicable results
Important outcomes considered
E. Lynch et al.
© 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Mortality outcomes were reported in two studies (44,67). Only one of these studies accounted for age and stroke severity in the calculations (44) and no significant association was found between mortality and timing of transfer to rehabilitation. The length of rehabilitation stay (RLOS) was observed in 9 studies. No significant relationship between RLOS and interval between stroke onset and transfer to rehabilitation was reported in 6 studies (43,47,48,53–55) (mean RLOS ranged from 43–78 days) whereas in 3 studies a significant direct relationship (61,63,67) (mean RLOS ranged from 30 days – 21 weeks) was observed.
Discussion From this review we have found that the current research is very limited regarding the ideal time point at which to commence physical rehabilitation and the benefits of commencing physical rehabilitation early. Further, the studies included in our review had several limitations including study design, small sample sizes and the ability to fully account for factors that may influence the associations with health outcomes. This review highlights that further well-designed research is needed. A summary of our findings and commentary relevant to our review questions is provided below. Effect of commencing physical rehabilitation within 7 days of stroke The majority of studies designed to evaluate early physical rehabilitation were focussed on early mobilization. Three studies provided evidence on the effects of mobilizing people within 24 h of stroke and clear clinical benefits over usual SU care were not demonstrated. The findings from our meta-analyses remain limited because the overall sample size was small with only 159 people. The question of the safety of mobilizing within 24 h remains unclear and we found a strong trend suggesting that mobilization within 24 h of stroke may increase the risk of dying within 3 months. The differences in study design and participant descriptors may explain the variable statistical heterogeneity (I2 ranged from 6%–66%) observed in the meta-analysis. The three RCTs included in the meta-analysis had minor variations in participant characteristics and SU mobilization protocols; for instance, the participants in one study (29) were on average younger than the other trials’ participants (11,30), and the SU mobilization protocols ranged from an average of 18 minutes/day (11) to several out of bed sessions per day (30). Combining an investigation of ‘early rehabilitation’ with ‘increased frequency of early mobilization’ may confound the results. The only RCT that kept intensity the same between groups (30) had a non-significant trend to worse outcomes when rehabilitation was commenced within 24 h. Significantly fewer complications were reported when people with stroke commenced physical rehabilitation 3 days rather than 7 days after stroke (27), and commencing physical rehabilitation by day 3 was significantly associated with better functional outcomes (35,36). Along with the consideration of function of the person with stroke, the pragmatics of early physical rehabilitation should be Vol ••, •• 2014, ••–••
5
Systematic review
E. Lynch et al.
Fig. 2 Meta-analysis of mortality within 3 months of stroke (odds ratio). CI, confidence interval; df, degrees of freedom.
Fig. 3 Meta-analysis of Barthel Index at 3 months post stroke (mean difference, fixed effects). CI, confidence interval; df, degrees of freedom; SD, standard deviation.
Fig. 4 Meta-analysis of having a good outcome (mRS 0–2) at 3 months post-stroke (odds ratio). CI, confidence interval; df, degrees of freedom.
Fig. 5 Meta-analysis of experiencing no complications in first 3 months post-stroke (odds ratio). CI, confidence interval; df, degrees of freedom.
considered. Only one of the studies included information on the cost-effectiveness of starting rehabilitation with 24 h of stroke and providing greater frequency of mobilization-related therapy sessions. Findings from this review are not strong enough to provide clear directives as to the specific time point at which a person should commence physical rehabilitation after stroke, other than that it appears physical rehabilitation should commence at some point within 3 days of stroke (27,35,36). Therefore, further ‘high quality’ research in this area is needed.
6
Vol ••, •• 2014, ••–••
Transferring to rehabilitation service Given the suggestion above that people with stroke should commence physical rehabilitation within three days of stroke, then physical rehabilitation should commence either in the SU or a transfer to the rehabilitation facility should occur within this timeframe. However, none of the included studies routinely transferred people with stroke to rehabilitation within the first week of stroke; people transferred within 1 week of stroke were generally outliers from the normal practice patterns. © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Appropriate method
Acceptable recruitment
© 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization x ✓ x x ✓ x ✓ x ✓ x x ✓ ✓ ✓ x ✓ ✓ x ✓
✓ ✓ x x x x ✓ ✓ ✓ ✓ ✓ x ✓ ✓ x ✓ x x ✓ x ✓ ✓ ✓ ✓ ✓ ✓ x ✓ ✓ x x ✓ ✓ x ✓ ✓ x ✓
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
x x ✓ ✓ ✓ ✓ x ✓ x x x ✓ ✓ x x x x x x x x x x x x x x x x x x x
x x x x x x x x ✓ x x x x
x ✓ x x
✓ ✓ ✓ ✓
x ✓ ✓ ✓
x ✓ ✓ x
Follow-up for 6/12
✓ x
Accounted for age, stroke severity
✓ ✓
✓ ✓
Outcome accurately measured
✓ ✓
Exposure accurately measured
NS: non-significant. Positive: results in favor of transferring to rehabilitation early for better functional outcomes.
Q1: Effect of commencing rehabilitation within 7 days Prospective observational cohort studies Horner et al. (32) ✓ ✓ ✓ Hu et al. (31) ✓ ✓ ✓ Retrospective observational cohort studies Hayes & Carroll (35) ✓ ✓ x Huang et al. (34) ✓ ✓ ✓ Kuwabara et al. (33) ✓ ✓ ✓ Matsui et al. (36) ✓ ✓ ✓ Q2: Effect of transferring to rehabilitation facility earlier Prospective observational cohort studies Ancheta et al. (38) ✓ ✓ x Anderson et al. (49) ✓ ✓ ✓ Bourestom (50) ✓ ✓ ✓ Horn et al. (39) ✓ ✓ ✓ Lin et al. 2003 (41) ✓ ✓ x Lin et al. 2000 (40) ✓ ✓ ✓ Massucci et al. (42) ✓ ✓ ✓ Maulden et al. (43) ✓ ✓ ✓ Musicco et al. (44) ✓ ✓ ✓ Paolucci et al. 2000 (45) ✓ ✓ ✓ Paolucci et al. 1998 (46) ✓ ✓ x Shah et al. (47) ✓ ✓ ✓ Suputtitada et al. (48) ✓ ✓ ✓ Retrospective observational cohort studies Bruell & Simon (51) ✓ ✓ x Chung et al. (52) ✓ ✓ x Feigenson et al. (a) (53) ✓ ✓ ✓ Feigenson et al. (b) (54) ✓ ✓ ✓ Gagnon et al. (55) ✓ ✓ ✓ Gowland (69) ✓ ✓ ✓ Inouye et al. (56) ✓ ✓ ✓ Johnston & Keister (57) ✓ ✓ ✓ Koh et al. (58) ✓ ✓ ✓ Kong et al. (59) ✓ ✓ x Novack et al. (60) ✓ ✓ ✓ Rossi et al. (61) ✓ ✓ ✓ Saeki et al. (62) ✓ ✓ ✓ Salter et al. (63) ✓ ✓ ✓ Stern et al. (64) ✓ ✓ x Tur et al. (65) ✓ ✓ ✓ Wang et al. (66) ✓ ✓ ✓ Wylie (67) ✓ ✓ ✓ Yavuzer et al. (68) ✓ ✓ ✓
Focussed issue
Table 3 Critical appraisal of all included cohort studies using Critical Appraisal Skills Programme (15)
positive NS positive positive NS positive positive NS positive positive positive positive positive positive positive positive positive positive positive
positive positive positive positive positive positive positive positive positive positive positive positive NS
positive NS NS positive
positive positive
What are results?
x x x x ✓ x x x ✓ x x x x x x x ✓ x x
✓ x x ✓ ✓ ✓ ✓ x x x ✓ x x
x x x ✓
✓ ✓
Precision of results
x x x x ✓ x ✓ x ✓ x x x x ✓ x ✓ ✓ x x
✓ x x ✓ ✓ ✓ ✓ ✓ ✓ ✓ x ✓ ✓
x x x ✓
✓ ✓
Believable results
unclear ✓ ✓ ✓ ✓ ✓ x x ✓ x x ✓ x ✓ x x ✓ x x
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ x x ✓ x
x x unclear ✓
x ✓
Applicable results
✓ x ✓ ✓ x ✓ ✓ x ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ x
✓ x x ✓
✓ ✓
Fit with other evidence
E. Lynch et al.
Systematic review
Vol ••, •• 2014, ••–••
7
Systematic review
E. Lynch et al.
Table 4 Summary data from studies investigating Question 2: What are the effects of earlier transfer to the rehabilitation service?
Country
n=
Stroke onset-transfer to rehabilitation interval (days)
Prospective observational studies Ancheta et al. (38) 2000
USA
87
0–14, 15–28, 29–42
Anderson et al. (49)
1974
USA
350
not presented
Bourestom (50) Horn et al. (39)
1967 2005
USA USA
53 830
Lin et al. (41)
2003
Taiwan
105
not presented mod: 11·4 ± 12·7, severe: 18·5 ± 29·5 within 90, mean 24·2
Lin et al. (40)
2000
Taiwan
110
within 60, mean 24·1
Massucci et al. (42) Maulden et al. (43)
2006 2005
Italy USA
997 969
26·1 ± 25·6 13·8 ± 18·7
Musicco et al. (44)
2003
Italy
1716
range up to >31
Paolucci et al. (45)
2000
Italy
145
Introduction Rehabilitation after stroke consists of individually tailored interventions which reduce stroke-related activity limitations and participation restrictions (1). An important issue being debated is when physical rehabilitation (defined as physical therapy, Correspondence: Elizabeth Lynch*, International Centre for Allied Health Evidence, School of Health Sciences, Sansom Institute for Health Research, University of South Australia, GPO Box 2471, Adelaide, SA 5001, Australia. E-mail: [email protected] 1 International Centre for Allied Health Evidence, School of Health Sciences, Sansom Institute for Health Research, University of South Australia, Adelaide, SA, Australia 2 Stroke Division, The Florey Institute of Neuroscience and Mental Health, Heidelberg, Vic., Australia 3 Stroke and Ageing Research Centre, Department of Medicine, Monash University, Clayton, Vic., Australia Received: 31 October 2013; Accepted: 16 December 2013 Conflict of interest: None declared. DOI: 10.1111/ijs.12262 © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
occupational therapy and physical exercises aimed at preserving or restoring physical function) (2) should commence after stroke. It has been postulated that physical rehabilitation commenced too early may increase the area of brain ischaemia and result in worse functional outcomes (3). Conversely, it has also been hypothesized that early physical rehabilitation may promote positive brain function reorganization and recovery (4), so a delay in rehabilitation may result in less effectiveness. In addition, it is well established that people with stroke are extremely inactive in hospital (5) and this increases the risk of losing muscle mass, strength and function (6). Therefore, a delay in commencing physical rehabilitation may result in the person with stroke developing more secondary impairments. Best practice guidelines are unanimous in recommending stroke unit (SU) care for people with stroke, with early rehabilitation being a defining component of a SU. However, the definition of ‘early’ rehabilitation is somewhat ambiguous. The Australian clinical guidelines for stroke recommend that rehabilitation should commence the first day after stroke (7), whereas European guidelines simply recommend rehabilitation should commence ‘early’ (8) and the United States guidelines recommend that rehabilitation should commence as early as possible, once medical stability has been attained (9). Mobilization is an important component of physical rehabilitation, and guidelines from the United Kingdom specify that mobilization should be commenced within 24 h of stroke (10). The recommendations in these international stroke guidelines have been based on one randomized controlled trial (RCT) (11), which was also the only study included in the 2009 Cochrane review (12). Different models of SUs exist whereby some are comprehensive stroke units, which are staffed to provide acute care as well as rehabilitation over a period of weeks (13,14) while others are acute stroke units, where patients are admitted acutely and discharged within approximately 3 to 10 days (13,14). Therefore, people admitted to acute SUs who require rehabilitation beyond this time point need to be transferred to a separate rehabilitation facility if inpatient rehabilitation is required. Knowing how soon a person with stroke should be transferred to a rehabilitation hospital if an acute hospital is unable to provide this therapy is important. The aims of this systematic review were to determine when it is best to commence physical rehabilitation post-stroke. Specifically: 1. To update: What are the effects on mortality, function and complications when physical rehabilitation is commenced ‘early’ i.e. within seven days of stroke? 2. When sustained rehabilitation is not provided by an acute SU, what are the effects of earlier transfer to a designated rehabilitation ward/hospital? Outcomes of interest in this review were differences in the health and functional outcomes of people with stroke. Vol ••, •• 2014, ••–••
1
Systematic review
E. Lynch et al.
Methods
Results
Data sources A computerized search was initially conducted by a single reviewer (EL) from inception to December 2012: databases Scopus, CINAHL, Medline, Embase, The Cochrane Library, and Pubmed were searched. The search strategy used a combination of the free text terms ‘stroke’ (sub-headings rehabilitation and therapy), ‘Cerebrovascular accident’ (sub-headings rehabilitation and therapy), ‘outcome assessment’, ‘time factors’, ‘outcomes health care’, ‘rehabilitation’ ‘commencement of rehabilitation’, ‘early ambulation’ and ‘early rehabilitation’.
The search yielded 3751 titles and abstracts; of which 48 articles were retained for the review (Fig. 1).
Study selection Any RCT, systematic reviews, prospective or retrospective controlled clinical trials or cohort studies published in English meeting the following criteria were included: • Examined people with stroke who received inpatient physical rehabilitation • Compared groups of people with stroke who commenced physical rehabilitation at different time intervals following stroke either in SUs or via transfer to rehabilitation units • Collected any objective outcome measures, for example mortality, complications, Functional Independence Measure (FIM), Barthel Index (BI), length of stay, discharge destination. Data collection and analysis One reviewer (EL) examined titles and abstracts against the inclusion criteria. Following removal of duplicates, full text articles were obtained for a more detailed review by two independent reviewers (EL and SH) to assess for inclusion. Reference lists were scrutinized for further relevant articles. In cases where further details were required, one reviewer (EL) contacted the articles’ authors. Data were extracted from papers that met the review criteria, and the risk of bias of each study was assessed using the Critical Appraisal Skills Program (CASP) criteria for RCTs and cohort study designs (15). One reviewer (EL) assessed each paper for risk of bias, a second reviewer (SH) assessed all RCTs and randomly assessed one third of the cohort studies to check for consistency. Where there was sufficient homogeneity of study design (RCT), population, intervention and measures, a meta-analysis was performed using the review software package Revman5 (16). This required the identification of the number of participants in each group for each trial, as well as the total number (for dichotomous data) and the mean and standard deviations for each group (for ordinal scale data). Data were analysed as odds ratios or mean differences using fixed effects with 95% confidence intervals. Where the means and standard deviations required to perform a meta-analysis were not published, these were requested from the authors. Only data from people who survived their stroke were included in the meta-analysis of functional independence, as mortality outcomes were analysed separately. Where metaanalysis was not possible, the results from the included papers were synthesized narratively.
2
Vol ••, •• 2014, ••–••
What are the effects of commencing physical rehabilitation within 7 days of stroke? Sixteen papers were included to answer this question (11,22–36) (Table 1). Six of the RCT papers had results from the same study cohort (11,22–26) so were analysed as one study. The included RCTs were generally of good quality with a low risk of bias (Table 2), however were based on small samples sizes and may have been underpowered to answer particular research questions related to outcomes. Commencing physical rehabilitation within 24 h Three RCTs used similar protocols to commence physical rehabilitation early on SUs, and compared this to usual SU care in Australia, United Kingdom and Norway (11,29,30). Overall, 159 people with stroke were included in early (versus usual SU care) rehabilitation trials. The main aim of these three studies was to assess the benefits of mobilizing patients within 24 h of stroke onset (11,29) or hospital admission (30) compared with usual care which was within 48 h of stroke. In two of the studies, the frequency of mobilization in the early rehabilitation group was increased (11,29). All studies provided outcomes at 3 months for case fatality, complications, functional independence using the BI and ‘good outcome’ indicated by a modified Rankin Score (mRS) of 0–2 (equivalent to none or slight disability). The comparator groups at baseline were mostly similar in terms of factors affecting post-stroke recovery such as age, comorbidities and stroke severity, and casemix adjustment was not conducted in any of these studies. We were able to conduct meta-analyses using these data and results of our meta-analyses are presented in Figs 2–5. We found a trend for fewer deaths in the usual care group which was mobilized 24–48 h post stroke [P = 0·06, OR 2·58 (95% CI 0·98– 6·79), Fig. 2]. At 3 months there was no significant difference between groups in BI [P = 0·23, OR 1·20 (95% CI – 0·77–3·18)], Figure 3. The likelihood of having a good outcome and experiencing no complications within 3 months following stroke was similar between groups, (P = 0·66 and P = 0·82 respectively, Figs 4 and 5). The mobilization sessions appeared to be well tolerated, since in the one study reporting blood pressure changes during mobilizations, no patient had 3 consecutive drops in blood pressure >30 mm Hg during the first 3 attempted mobilizations (11). Commencing physical rehabilitation within first 7 days of stroke (Table 1) One RCT provided evidence that commencing physical rehabilitation 3 days compared to 7 days after stroke resulted in significantly fewer serious complications, and did not influence cerebral blood flow (27). Authors of a previously conducted RCT which compared SU care to general ward care (37) conducted a follow-up analysis and reported that after adjusting for age and severity of stroke, the time at which physical rehabilitation commenced (on day of admission on SU vs day 2 or 3 on medical ward) was an independent factor significantly associated with discharge home within 6 weeks of stroke (28). © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Systematic review
E. Lynch et al.
Records iden fied through database searching (n = 3751)
Records a er duplicates removed (n = 3127)
Records screened (n = 3127)
Addi onal relevant studies iden fied through hand searching of literature (n = 12)
Full-text ar cles assessed for eligibility (n = 87)
Ar cles included in qualita ve synthesis (n = 48, 43studies) Prospec ve observa onal cohort studies (n = 15) Retrospec ve observa onal cohort studies (n = 23) Randomised controlled trials (n = 10, 5 studies)
Records excluded (n = 3052)
Full-text ar cles excluded = 32 1. Study design and repor ng • No comparison of groups (n = 4) • No comparison/record regarding stroke onset-rehabilita on commencement interval (n = 6) • Comparing different models of care, effect of ming not analyzed (n = 7) • Review ar cle where ming of rehabilita on not main focus (n = 9) • Func onal outcomes not collected (n = 1) 2. Popula on • Not first rehabilita on admission for current stroke (n = 1) • Not inpa ent physical rehabilita on (n = 4) + Eligible ar cles not included = 7 1. Presented same data as another ar cle included in review (14, 17–21) (n = 6) 2. Review ar cle without systema c search (3) (n = 1)
Studies included in quan ta ve synthesis (meta-analysis) (n = 3)
Fig. 1 Process for identification of the included studies.
Six cohort studies (31–36) variously described the effect of timing of physical rehabilitation commencement on patient outcomes. Three of 4 cohort studies which used regression analyses to evaluate the effect of timing of rehabilitation commencement on BI, provided evidence that a better BI score on discharge was significantly associated with the timing of physical rehabilitation initiation (smaller delay, greater BI) (31–34). Authors of two cohort studies analysed the effects of commencing physical rehabilitation within 3 days (compared to 4 or more days), and reported that commencing within 3 days had no significant association with mortality (36), improved the likelihood of reducing disability by 15·3% (P < 0·001) (36); and resulted in better ambulation and ADL function, rated via non-validated outcome measurement tools (35). What are the effects of earlier transfer to the rehabilitation service? Thirteen prospective (38–50) and 19 retrospective observational cohort studies (51–69) met the inclusion criteria for this question. © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
In general, the quality of the studies regarding the effects of earlier transfer to a rehabilitation service after stroke was poor with a high risk of bias identified in a number of the included studies (Table 3). Studies were highly heterogeneous, with authors using different reporting methods, outcome measures and analyses (Table 4). The average time at which people with stroke were transferred to rehabilitation ranged from 11 days (39) to 75 days post-stroke (60). All 32 studies had some measure of clinical outcome, the most common being FIM (16 studies) or BI (7 studies). Twentysix of the 32 included studies (81%) accounted for age and stroke severity in the calculations relating functional outcome with the timing of transfer to rehabilitation. Among these 26 studies, the majority (n = 23, 88%) had results that favored early transfer to rehabilitation with regards to better functional outcome. The optimal time to transfer to the rehabilitation service was calculated in one study (61) and transfer within 12 days of stroke was recommended for best outcome based on a trend analysis of LOS efficiency (change in FIM divided by LOS). Vol ••, •• 2014, ••–••
3
4
Vol ••, •• 2014, ••–••
2010
2012
2011
1999 1991
2003
2010
2009
1986
2011
2010
Langhorne et al. (29)
Sundseth et al. (30)
Diserens et al. (27)
Indredavik et al. (28) Indredavik et al. (37)
Horner et al. (32)
Hu et al. (31)
Huang et al. (34)
Hayes & Carroll (35)
Kuwabara et al. (33)
Matsui et al. (36)
Japan
Japan
USA
Taiwan
Taiwan
USA
Norway
Netherlands
Norway
5482
45014
30
76
154
738
206
42
56
32
71
Australia
2008 2009 2008 2010 2010 2008
Bernhardt et al. (11) Sorbello et al. (24) Cumming et al. (22) Cumming et al. (23) Tyedin et al. (26) Tay-Teo et al. (25)
United Kingdom
n=
Country
Date
Author
Retrospective cohort study Retrospective cohort study
Post-hoc analysis from single centre RCT Prospective cohort study Prospective cohort study Retrospective cohort study Retrospective cohort study
Single centre RCT
Single centre RCT
Single centre RCT
Multi-centre RCT, Phase II
Study type
In-hospital mortality mRS ≤ 1 at discharge
3 days (delayed)
Not presented
LOS ADL Gait ability BI
BI at 3 months
BI at discharge
No complications in hospital mRS (0–2) PSV at 72–96 h EDV at 72–96 h Factors associated with discharge home within 6 weeks of stroke BI at 1 year
Mixed model multiple regression: Estimate −1·79, SE 0·55, P < 0·001 Multiple linear regression: β = −0·65, 95% CI −1·2 to −0·10, adjusted R2 = 0·64, P = 0·02 Multivariate analysis by stepwise linear regression: β = −2·45, SE = 0·5, P < 0·01 early 24 ± 12 days vs delayed 39 ± 15, P* = 0·005 early 2·40 ± 0·83 vs delayed 2·13 ± 0·74 early 2·87 ± 1·30 vs delayed 2·73 ± 1·30 Not significant: One day delay in initiating rehabilitation: B = 0·003, 95% CI −0·042 to 0·047 Early 1·7% vs delayed 1·6%, univariate P = 0·799 Favors commencing early: Marginal effect 0·048, 95% CI 0·013 to 0·084.
EM 7/27 (25·9%) vs SC 2/29 (6·9%), OR 4·73; 95% CI, 0·89–25·21; P = 0·07 EM 10/25 (40%) vs SC 17/28 (60·7%), OR 2·32, 95% CI, 0·77–6·98; P = 0·14 EM 9/27 (33·3%) vs SC 10/29 (24·4%); P = 0·93 EM median 19, IQR 7–20, SC median 19, IQR 6–20 Mean (SD) EM 3·9 (3·8), SC 5·5 (2·9), P = 0·02 3 days 18/25 (72%) vs 7 days 9/17 (53%) P* = 0·33 3 days 10/25 (40%) vs 7 days 6/17 (35%) P* = 1·00 3 days 75 ± 6 vs 7 days 73 ± 6 3 days 30 ± 3 vs 7 days 30 ± 3·5 Time to start of mobilization, coefficient −0·4862, relative risk 0·62, CI 0·45–0·84
Death mRS (0–2) No complications BI Improvement in NIHSS
Death mRS (0–2) No complications BI
EM 8/38 vs SC 3/33; absolute risk difference = 12%; 95% CI −4·3% to 28·2%, P = 0·20 EM 15/38 (39·5%) vs SC 10/33 (30·3%), P 0·46. EM 5/38 vs SC 8/33, P* = 0·36 EM median 16·5, IQR 2·0 to 20·0, SC median 16·5, IQR 9·0 to 20·0, P = 0·713 EM 10 (5–17) vs SC 12 (4–15), z = −0·057, P = 0·755 EM median 0·32 vs SC 0·24, P = 0·17 EM AUD 13 559 ± 12 667 vs SC AUD 21 860 ± 16 272, P = 0·02 EM 0/16 vs SC 1/16 EM 12/16 (75%) vs SC 7/16 (43·8%), P 0·07 EM 7/16 (43·8%) vs SC 4/16 (25%), P* = 0·46 EM median 20, IQR 18–20, SC median 17, IQR 2–20, P = 0·21
results
Death mRS (0–2) No complications BI IDA scale AQoL at 12 months Total cost
outcomes (at 3 months unless otherwise stated)
3 days (delayed)
Mean 6·7 ± 6·7 days (range 1–50 days) Mean 7·7 ± 4·6 days
Day of admission (SU) vs day 2 or 3 (general ward) Median 3 days
At 3 days (early) vs 7 days (delayed)
Within 24 h of stroke vs 24–48 h. Also increased frequency of mobilization in EM group Within 24 h of hospital admission vs 24–48 h
Within 24 h of stroke vs 24–48 h. Also increased frequency of mobilization in EM group
Time rehabilitation commenced
Table 1 Summary date from studies investigating Question 1: What are the effects of commencing physical rehabilitation within 7 days of stroke?
Systematic review E. Lynch et al.
© 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Systematic review
✓ ✓ ✓ x ✓ ✓ ✓ ✓ ✓ x ✓ ✓ ✓ ✓ unclear ✓ ✓ ✓ ✓ ✓ Bernhardt 2008 (11) Langhorne 2010 (29) Indredavik 1999 (28) Sundseth 2012 (30) Diserens 2011 (27)
✓ ✓ y ✓ ✓
✓ ✓ ✓ ✓ ✓
✓ ✓ s ✓ ✓
✓ ✓ ✓ x ✓
✓ ✓ ✓ ✓ ✓
Small Small Not applicable nil small
✓ s ✓ ✓ ✓
Benefit worth harms & costs Precision of results Treatment effect size Same management (aside from intervention) Similar at baseline Assessors blinded Patients accounted for Random assignment Focussed issue
Table 2 Critical appraisals of included randomized controlled trials using Critical Appraisal Skills Programme criteria (15)
Locally applicable results
Important outcomes considered
E. Lynch et al.
© 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Mortality outcomes were reported in two studies (44,67). Only one of these studies accounted for age and stroke severity in the calculations (44) and no significant association was found between mortality and timing of transfer to rehabilitation. The length of rehabilitation stay (RLOS) was observed in 9 studies. No significant relationship between RLOS and interval between stroke onset and transfer to rehabilitation was reported in 6 studies (43,47,48,53–55) (mean RLOS ranged from 43–78 days) whereas in 3 studies a significant direct relationship (61,63,67) (mean RLOS ranged from 30 days – 21 weeks) was observed.
Discussion From this review we have found that the current research is very limited regarding the ideal time point at which to commence physical rehabilitation and the benefits of commencing physical rehabilitation early. Further, the studies included in our review had several limitations including study design, small sample sizes and the ability to fully account for factors that may influence the associations with health outcomes. This review highlights that further well-designed research is needed. A summary of our findings and commentary relevant to our review questions is provided below. Effect of commencing physical rehabilitation within 7 days of stroke The majority of studies designed to evaluate early physical rehabilitation were focussed on early mobilization. Three studies provided evidence on the effects of mobilizing people within 24 h of stroke and clear clinical benefits over usual SU care were not demonstrated. The findings from our meta-analyses remain limited because the overall sample size was small with only 159 people. The question of the safety of mobilizing within 24 h remains unclear and we found a strong trend suggesting that mobilization within 24 h of stroke may increase the risk of dying within 3 months. The differences in study design and participant descriptors may explain the variable statistical heterogeneity (I2 ranged from 6%–66%) observed in the meta-analysis. The three RCTs included in the meta-analysis had minor variations in participant characteristics and SU mobilization protocols; for instance, the participants in one study (29) were on average younger than the other trials’ participants (11,30), and the SU mobilization protocols ranged from an average of 18 minutes/day (11) to several out of bed sessions per day (30). Combining an investigation of ‘early rehabilitation’ with ‘increased frequency of early mobilization’ may confound the results. The only RCT that kept intensity the same between groups (30) had a non-significant trend to worse outcomes when rehabilitation was commenced within 24 h. Significantly fewer complications were reported when people with stroke commenced physical rehabilitation 3 days rather than 7 days after stroke (27), and commencing physical rehabilitation by day 3 was significantly associated with better functional outcomes (35,36). Along with the consideration of function of the person with stroke, the pragmatics of early physical rehabilitation should be Vol ••, •• 2014, ••–••
5
Systematic review
E. Lynch et al.
Fig. 2 Meta-analysis of mortality within 3 months of stroke (odds ratio). CI, confidence interval; df, degrees of freedom.
Fig. 3 Meta-analysis of Barthel Index at 3 months post stroke (mean difference, fixed effects). CI, confidence interval; df, degrees of freedom; SD, standard deviation.
Fig. 4 Meta-analysis of having a good outcome (mRS 0–2) at 3 months post-stroke (odds ratio). CI, confidence interval; df, degrees of freedom.
Fig. 5 Meta-analysis of experiencing no complications in first 3 months post-stroke (odds ratio). CI, confidence interval; df, degrees of freedom.
considered. Only one of the studies included information on the cost-effectiveness of starting rehabilitation with 24 h of stroke and providing greater frequency of mobilization-related therapy sessions. Findings from this review are not strong enough to provide clear directives as to the specific time point at which a person should commence physical rehabilitation after stroke, other than that it appears physical rehabilitation should commence at some point within 3 days of stroke (27,35,36). Therefore, further ‘high quality’ research in this area is needed.
6
Vol ••, •• 2014, ••–••
Transferring to rehabilitation service Given the suggestion above that people with stroke should commence physical rehabilitation within three days of stroke, then physical rehabilitation should commence either in the SU or a transfer to the rehabilitation facility should occur within this timeframe. However, none of the included studies routinely transferred people with stroke to rehabilitation within the first week of stroke; people transferred within 1 week of stroke were generally outliers from the normal practice patterns. © 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization
Appropriate method
Acceptable recruitment
© 2014 The Authors. International Journal of Stroke © 2014 World Stroke Organization x ✓ x x ✓ x ✓ x ✓ x x ✓ ✓ ✓ x ✓ ✓ x ✓
✓ ✓ x x x x ✓ ✓ ✓ ✓ ✓ x ✓ ✓ x ✓ x x ✓ x ✓ ✓ ✓ ✓ ✓ ✓ x ✓ ✓ x x ✓ ✓ x ✓ ✓ x ✓
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
x x ✓ ✓ ✓ ✓ x ✓ x x x ✓ ✓ x x x x x x x x x x x x x x x x x x x
x x x x x x x x ✓ x x x x
x ✓ x x
✓ ✓ ✓ ✓
x ✓ ✓ ✓
x ✓ ✓ x
Follow-up for 6/12
✓ x
Accounted for age, stroke severity
✓ ✓
✓ ✓
Outcome accurately measured
✓ ✓
Exposure accurately measured
NS: non-significant. Positive: results in favor of transferring to rehabilitation early for better functional outcomes.
Q1: Effect of commencing rehabilitation within 7 days Prospective observational cohort studies Horner et al. (32) ✓ ✓ ✓ Hu et al. (31) ✓ ✓ ✓ Retrospective observational cohort studies Hayes & Carroll (35) ✓ ✓ x Huang et al. (34) ✓ ✓ ✓ Kuwabara et al. (33) ✓ ✓ ✓ Matsui et al. (36) ✓ ✓ ✓ Q2: Effect of transferring to rehabilitation facility earlier Prospective observational cohort studies Ancheta et al. (38) ✓ ✓ x Anderson et al. (49) ✓ ✓ ✓ Bourestom (50) ✓ ✓ ✓ Horn et al. (39) ✓ ✓ ✓ Lin et al. 2003 (41) ✓ ✓ x Lin et al. 2000 (40) ✓ ✓ ✓ Massucci et al. (42) ✓ ✓ ✓ Maulden et al. (43) ✓ ✓ ✓ Musicco et al. (44) ✓ ✓ ✓ Paolucci et al. 2000 (45) ✓ ✓ ✓ Paolucci et al. 1998 (46) ✓ ✓ x Shah et al. (47) ✓ ✓ ✓ Suputtitada et al. (48) ✓ ✓ ✓ Retrospective observational cohort studies Bruell & Simon (51) ✓ ✓ x Chung et al. (52) ✓ ✓ x Feigenson et al. (a) (53) ✓ ✓ ✓ Feigenson et al. (b) (54) ✓ ✓ ✓ Gagnon et al. (55) ✓ ✓ ✓ Gowland (69) ✓ ✓ ✓ Inouye et al. (56) ✓ ✓ ✓ Johnston & Keister (57) ✓ ✓ ✓ Koh et al. (58) ✓ ✓ ✓ Kong et al. (59) ✓ ✓ x Novack et al. (60) ✓ ✓ ✓ Rossi et al. (61) ✓ ✓ ✓ Saeki et al. (62) ✓ ✓ ✓ Salter et al. (63) ✓ ✓ ✓ Stern et al. (64) ✓ ✓ x Tur et al. (65) ✓ ✓ ✓ Wang et al. (66) ✓ ✓ ✓ Wylie (67) ✓ ✓ ✓ Yavuzer et al. (68) ✓ ✓ ✓
Focussed issue
Table 3 Critical appraisal of all included cohort studies using Critical Appraisal Skills Programme (15)
positive NS positive positive NS positive positive NS positive positive positive positive positive positive positive positive positive positive positive
positive positive positive positive positive positive positive positive positive positive positive positive NS
positive NS NS positive
positive positive
What are results?
x x x x ✓ x x x ✓ x x x x x x x ✓ x x
✓ x x ✓ ✓ ✓ ✓ x x x ✓ x x
x x x ✓
✓ ✓
Precision of results
x x x x ✓ x ✓ x ✓ x x x x ✓ x ✓ ✓ x x
✓ x x ✓ ✓ ✓ ✓ ✓ ✓ ✓ x ✓ ✓
x x x ✓
✓ ✓
Believable results
unclear ✓ ✓ ✓ ✓ ✓ x x ✓ x x ✓ x ✓ x x ✓ x x
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ x x ✓ x
x x unclear ✓
x ✓
Applicable results
✓ x ✓ ✓ x ✓ ✓ x ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ x
✓ x x ✓
✓ ✓
Fit with other evidence
E. Lynch et al.
Systematic review
Vol ••, •• 2014, ••–••
7
Systematic review
E. Lynch et al.
Table 4 Summary data from studies investigating Question 2: What are the effects of earlier transfer to the rehabilitation service?
Country
n=
Stroke onset-transfer to rehabilitation interval (days)
Prospective observational studies Ancheta et al. (38) 2000
USA
87
0–14, 15–28, 29–42
Anderson et al. (49)
1974
USA
350
not presented
Bourestom (50) Horn et al. (39)
1967 2005
USA USA
53 830
Lin et al. (41)
2003
Taiwan
105
not presented mod: 11·4 ± 12·7, severe: 18·5 ± 29·5 within 90, mean 24·2
Lin et al. (40)
2000
Taiwan
110
within 60, mean 24·1
Massucci et al. (42) Maulden et al. (43)
2006 2005
Italy USA
997 969
26·1 ± 25·6 13·8 ± 18·7
Musicco et al. (44)
2003
Italy
1716
range up to >31
Paolucci et al. (45)
2000
Italy
145
