What is the Probability of Patients who are Nonambulatory after Stroke Regaining Independent Walking? a Systematic Review

Abstract

Patients after stroke who are nonambulatory require resources, and independent walking becomes a major determinant of the ability to participate in activities of daily living. Our objective was to determine the probability of walking for patients who are nonambulatory in the first month after stroke. We performed a systematic review and meta-analysis of consecutive, prospective studies of nonambulatory patients within the first month after stroke in rehabilitation and acute units. The outcomes were the probability of achieving independent walking at three-, six- and 12 months after stroke. Twenty-six studies were included in the review. Seventeen studies comprising 2856 participants were entered into meta-analyses. For initially nonambulatory stroke patients managed in a rehabilitation unit, the probability of independent walking was 0·60 (95% CI 0·47–0·74, 1373 participants) at three-months, 0·65 (95% CI 0·53–0·77, 444 participants) at six-months and 0·91 (95% CI 0·81–1·00, 24 participants) at 12 months. For patients managed in an acute unit, the probability of independent walking was 0·39 (95% CI 0·27–0·52, 634 participants) at three-months, 0·69 (95% CI 0·46–0·92, 405 participants) at six-months and 0·74 (95% CI 0·59–0·88, 34 participants) at 12 months. 60% of patients managed in a rehabilitation unit who are nonambulatory in the first month after stroke will regain independent walking compared with 39% of those managed in an acute unit. This information can be used clinically to make decisions about allocation of rehabilitation resources, education of patients and carers, and for discharge planning.

Keywords

gait locomotion meta-analysis outcome prognosis rehabilitation stroke systematic review walking

Introduction

The ability to regain independent walking after stroke is of great importance to patients and health care providers alike. People fear losing independence more than anything else and the ability to walk promotes independence (1). Often after stroke, the probability of walking again is the first question that patients ask of health care providers. Regaining the ability to walk is a major determinant of a person's ability to participate in activities of daily living and therefore of discharge destination (2, 3).

There are two systematic reviews evaluating prognostic factors relating to walking after stroke (4, 5). Kwakkel and colleagues' review suggested that the predictors for outcome of walking after stroke were age; previous stroke; urinary continence; consciousness at onset; disorientation in time and place; severity of paralysis; sitting balance; admission ADL score and level of social support and metabolic rate of glucose outside the infarct area in hypertensive patients (4). More recently, Meijer and colleagues' review suggested that there was insufficient quality of data to provide predictions, although there was some evidence that low initial ADL functioning, high age, cognitive disturbance, paresis of the limbs, reduced initial level of consciousness, previous hemiplegia, homonymous hemianopia, visual extinction, constructional apraxia, no admission to a stroke unit, nonlacunar stroke, visuospatial construction problems, urinary incontinence and female gender were related to poorer outcome after stroke (5). Both these reviews included patients who were initially ambulatory and nonambulatory. What is still not clear is the probability of independent walking for people who are nonambulatory initially after stroke. Patients who are nonambulatory have generally had a devastating stroke, and patients with severe strokes require substantial physical and economic resources for rehabilitation or future care (6, 7). Thus, the objective of this systematic review was to determine the probability of walking for patients who are nonambulatory in the first month after stroke.

This systematic review examined patients managed in rehabilitation and acute units separately, because it is possible that there will be different outcomes in terms of walking between these settings. For example, acute units provide care for a full range of patients after stroke, with varying degrees of disability, including patients who are likely to walk quickly and go directly home without requiring inpatient rehabilitation. Therapeutic input may be more extensive in some rehabilitation units, thus also contributing to a different walking outcome. We were interested in examining the probability of regaining independent walking because it is important in enabling patients to participate in the community (8). Given that the majority of functional recovery after stroke occurs within the first 12 months (9) we examined the recovery of walking across this time frame, which encompasses inpatient rehabilitation (average discharge about three-months) and outpatient rehabilitation (average discharge about six-months).

The specific research questions for this study were:

In patients, who are managed in a rehabilitation unit and are nonambulatory within one-month of stroke, what is the probability of regaining independent walking within 12 months?

Is the probability any different for patients who are managed in an acute unit?

Knowing the probability of regaining walking of patients who are initially nonambulatory will provide a benchmark for walking outcomes. This information may also contribute to the process of triaging patients in acute units to home, rehabilitation or a care facility and would also allow education of family and carers regarding general expectations of walking outcome in patients who are nonambulatory in the first month after stroke.

Method

Identification and selection of studies

Searches were conducted of the following databases: MEDLINE (1966 to September 2010), CINAHL (1982 to September 2010), EMBASE (1980 to September 2010), Web of Science (1982 to September 2010) and Scopus (1960 to September 2010), without language restrictions, for relevant articles. Search terms included words relating to stroke (e.g. cerebrovascular disorder, CVA, infarct, bleed, brain, hemiplegia, etc.), prognosis (e.g. risk, outcome assessment, predict, follow-up, etc.), and locomotion (e.g. walking, gait, ambulation, mobility, etc.). Title and abstracts were displayed and screened by one reviewer to identify relevant studies. Full paper copies of relevant studies were retrieved and their reference lists were screened. The methods of retrieved papers were examined against predetermined inclusion criteria (Fig. 1) by two independent reviewers. Conflict of opinion was resolved by consensus after discussion with a third reviewer.

Fig. 1

Inclusion criteria.

Assessment of characteristics of studies

Quality

The quality of included studies was determined using the screening tool of prognostic studies developed by Hayden et al. (10). Studies were evaluated using the four categories relevant to this systematic review: study participation (criteria included that the sample represented the population of interest in key characteristics), study attrition (criteria included that the loss to follow-up was not associated with key characteristics), measurement of outcomes (criteria included that the measurement was clearly defined as well as valid and reliable) and measurement of confounders (criteria included that the important potential confounders were appropriately accounted for by valid and reliable measurement). Studies were rated as either meeting the criteria, partly meeting the criteria or not meeting the criteria for each of these categories. A study was defined as good quality if it either

met the criteria for three categories or

met the criteria for two categories and partly met the criteria for the other two categories.

Studies were evaluated by two reviewers independently with disagreements resolved by a third reviewer.

Participants

Participants had to be patients within one-month after a stroke, who were nonambulatory. Age, gender, the number of participants in the sample, inclusion criteria into the study and the time between stroke onset and initial walking measure were recorded and compared to examine similarity between the studies.

Settings

Studies where participants were admitted to hospital were included. The setting where the initial measurement of walking ability was completed was recorded and compared to examine similarity between the studies. Acute units included acute stroke units or general medical units, thereby comprising a nonselected population of patients after stroke. Rehabilitation units included any inpatient rehabilitation unit, where patients were selected and transferred after an acute episode of care.

Outcome measures

Measures of walking, time of measurement since stroke and definition of nonambulatory, as well as definition of ambulatory, were recorded and compared to examine similarity between the studies. Where multiple measures of walking were presented, independent walking was defined as being able to walk with/without aids, but without human assistance.

Data analysis

Data were extracted from the included studies by one reviewer and cross-checked by a second reviewer. Information about the method (i.e. design, participants, measures) and outcome data (i.e. number of participants that could walk independently) was extracted. Authors were contacted where there was difficulty extracting the data.

Walking outcome was examined over a 12-month time frame, three categories were then developed: three-, six- and 12-months after stroke. Meta-analyses were performed to calculate the pooled estimate of the proportion of patients that could walk independently and reported with 95% CI. A fixed effects model was used. In the case of significant statistical heterogeneity, I²>50% (11), a random effects model was applied to check the robustness of the results. The meta-analyses were performed using The MIX–Meta-Analysis Made Easy program version 1.7 (12, 13).

Results

Flow of studies through the review

The search returned 1010 studies. After screening the titles and abstracts, 66 papers were retrieved for evaluation of full text. Forty studies failed to meet the inclusion criteria and therefore 26 papers were included in the review. Of these 26 papers, after contacting authors, 17 had data available to be entered into meta-analyses (Fig. 2).

Fig. 2

Flow of studies through the review. *Papers may not have been included for failing to meet more than one inclusion criteria.

Description of studies

The quality of the included studies is outlined in Table 1 and a summary of the studies is presented in Table 2.

Table 1

Quality of included studies (n=26)

Study	Study participation – sample represents the population of interest in key characteristics	Study attrition – loss to follow-up is not associated with key characteristics	Outcome measurement if adequate	Important potential confounders are appropriately accounted for
Bacciglieri et al. (14)	Y	Y	Y	P
Baer and Smith (15)	Y	Y	Y	N
Bagg et al. (16)	Y	Y	Y	P
Daviet et al. (17)	N	N	Y	P
De Wit et al. (18)	Y	Y	Y	P
Ekstrand et al. (38)	P	Y	Y	P
Friedman (19)	Y	Y	Y	P
Jorgensen et al. (20)	Y	Y	Y	P
Kuys et al. (21)	Y	Y	Y	P
Loewen and Anderson (22)	Y	Y	Y	P
Macciocchi et al. (23)	P	Y	Y	P
Minelli et al. (24)	Y	Y	Y	N
Olsen (25)	P	Y	Y	N
Paolucci et al. (26)	Y	Y	Y	P
Patel et al. (27)	Y	Y	Y	P
Petrilli et al. (28)	P	Y	Y	P
Piron et al. (29)	P	Y	Y	P
Prevo et al. (30)	P	Y	Y	P
Reid et al. (31)	P	Y	P	P
Singh et al. (32)	Y	Y	Y	N
Sommerfeld and von Arbin (2)	P	Y	Y	P
Thommessen et al. (33)	P	Y	Y	P
Van Bennekom et al. (34)	Y	Y	Y	Y
Viosca et al. (35)	Y	N	Y	P
Wade et al. (36)	Y	N	Y	N
Ween et al. (37)	Y	Y	Y	P

Y, yes; N, no; P, partly.

Table 2

Summary of included studies (n=26)

Study	Participants		Setting	Measurement of walking
	Characteristics	Inclusion		Initial			Follow-up
	Characteristics	Inclusion		Measurement	Definition of nonambulatory	Time (days)	Measurement	Definition of ambulatory	Time
Bacciglieri et al. (14)	n=33/33 Age (years)=64 (SD 7)	Stroke	Rehab	Three-point custom made scale	With assistance (human) Custom scale score 1, 2	14	Three-point custom made scale	Without assistance (human) Custom scale score 3	D/C
Baer and Smith (15)	n=unknown/238 Age (years)=72 (SD 11) Gender=92 M, 93 F	Stroke	Rehab	10 m walk test	With assistance (human)	3	10 m walk test	Without assistance (human)	D/C
Bagg et al. (16)	n=561 Age (years)=71 (SD 12) Gender=302 M, 259 F	1st stroke, acutely stable, no severe cognitive impairment	Rehab	FIM – mobility subscale	With assistance (human) FIM 1–5	31	FIM – mobility subscale	Without assistance (human) FIM 6, 7	D/C (three-months)
Daviet (17)	n=unknown/156	1st stroke, no SAH	Rehab	BI	With assistance (human) BI mobility score 0, 5, 10	2	BI	Without assistance (human) BI mobility score 15	12 months
De Wit et al. (18)	n=388/532 Age (years)=69 Gender=283 M, 249 F	1st stroke, score on RMA gross function <11; leg/trunk <8; arm <12	Rehab	BI	With assistance (human) BI mobility score 0, 5, 10	19	BI	Without assistance (human) BI mobility score 15	Two-, four-, six-months
Ekstrand et al. (38)	n=46/60 Age (years)=66 Gender=34 M, 26 F	1st stroke, motor deficits at four-day poststroke, no aphasia	Acute	COVS	With assistance (human)	2	COVS	Without assistance (human)	Three-months
Friedman (19)	n=113/113 Age (years)=74 (SD 6) Gender 62 M, 51 F	Stroke, >60 years	Acute	Walking speed	With assistance (human)	7	Walking speed	Walking speed >0·150 m/s	Three-, four-months
Jorgensen et al. (20)	n=510/804 Age (years)=74 (SD 11)	Stroke, no aphasia, no disorientation	Acute	BI	Immobile or walks <50 yd	7	BI	Walks >50 yd without assistance (human)	D/C (one-month), three–months
Kuys et al. (21)	n=58/105 Age (years)=70 (SD 13) Gender=64 M, 41F	Stroke	Rehab	MAS Item 5	With assistance (human) MAS Item 5 score 0–2	16	MAS Item 5	Without assistance (human) MAS Item 5 score 3–6	D/C (two-,·four-months)
Loewen and Anderson (22)	n=unknown/50 Age (years)=68 (SD 10) Gender=28 M, 22 F	Stroke, no altered consciousness	Acute	MAS Item 5	With assistance (human) MAS Item 5 score 0–2	7	MAS Item 5	Without assistance (human) MAS Item 5 score 3–6	D/C (two-months)
Macciocchi et al. (23)	n=unknown/328 Age (years)=70 (SD 10) Gender=179 M, 149 F	MCA ischemic stroke	Acute	BI	With assistance (human) BI mobility score 0, 5, 10	10	BI	Without assistance (human) BI mobility score 15	Three-months
Minelli et al. (24)	n=34/81 Age (years)=65 (SD 12) Gender=51 M, 30 F	Stroke	Acute	BI	With assistance (human) BI mobility score 0, 5, 10	2	BI	Without assistance (human) BI mobility score 15	12 months
Olsen (25)	n=72/72 Age (years)=68 (SD 10) Gender=39 M, 33 F	Stroke, nonambulatory	Rehab	Five-point custom made scale	With assistance (human) Custom scale score 1–4	28	Five-point custom made scale	Without assistance (human) Custom scale score 5	Three-months
Paolucci et al. (26)	n=437/437 Age (years)=68 (SD 13) Gender=260 M, 240 F	Ischemic stroke	Rehab	Custom made scale	Not walking or in wheelchair	21	Custom made scale	Without assistance (human)	D/C (two-,·five-months)
Patel et al. (27)	n=322/459 Age (years)=70 (SD 11) Gender=260 M, 240 F	Stroke	Acute	FIM mobility subscale	With standby assistance (human) FIM 1–5	2	FIM mobility subscale	Without assistance (human) FIM 6, 7	Three-, six-months
Petrilli et al. (28)	n=93/93 Age (years)=65 Gender 46 M, 47 F	1st stroke	Rehab	Custom made scale	10 m with assistance (human)	18	Custom made scale	10 m without assistance (human)	D/C (two-months)
Piron et al. (29)	n=13/20 Age (years)=68 (SD 6) Gender=8 M, 12 F	1st stroke, nonambulatory, dep for ADLs	Rehab	Hemiplegic stroke scale (HSS)	With assistance (human) HSS score >3	30	HSS	Without assistance (human) HSS score <4	Seven-months
Prevo et al. (30)	n=43/45 Age (years)=58 Gender=21 M, 22 F	1st stroke, nonambulatory, dep for ADLs	Rehab	Three-point custom made scale	With assistance (human) or wheelchair dependent	30	Three-point custom made scale	Without assistance (human)	Six-months
Reid et al. (31)	n=292/538 Age (years)=74 Gender 286 M, 252 F	Stroke	Acute	Custom made scale	With assistance (human)	1	Modified Rankin scale	Without assistance (human) mRS score 0–3	Six-months
Singh et al. (32)	n=255/255 Age (years)=57 (SD 9) Gender=147 M, 108 F	Stroke, nonambulatory	Rehab	FIM	With assistance (human) FIM 1–4	17	FIM	Standby assistance (human) FIM 5–7	D/C (three-months)
Sommerfeld and von Arbin (2)	n=98/115 Age (years)=80 (SD 7) Gender=48 M, 67 F	Stroke, >65 years, no severe confusion, no aphasia	Acute	BI	With assistance (human) BI mobility score 0, 5, 10	10	BI	Without assistance (human) BI Mobility score 15	D/C (three-months)
Thommessen et al. (33)	n=unknown/171 Age (years)=78 (SD 11) Gender=69 M, 102 F	Stroke >60 years, no SAH	Rehab	BI	With assistance (human) BI mobility score 0, 1, 2	10	BI	Without assistance (human) BI mobility score 3	D/C (two-months), 12 months
Van Bennekom et al. (34)	n=unknown/125 Age (years)=73 (SD 11) Gender=67 M, 58 F	Stroke, no SAH	Acute	BI	With assistance (human) BI mobility score 0, 1, 2	14	BI	Without assistance (human) BI mobility score 3	Three-, six-months
Viosca et al. (35)	n=26/26 Age (years)=70 (SD 9) Gender=15 M, 11 F	Stroke, nonambulatory, able to follow instructions, sitting for 5 s	Rehab	FACHS	With assistance (human) FACHS 0, 1	11	FACHS	Without assistance (human) FACHS 2–5	D/C (three-months). 12 months
Wade et al. (36)	n=39/101 Age (years)=68 Gender=23 M, 78 F	Stroke	Acute	Five-point custom made scale	With assistance (human) Custom scale score 3–5	5	Five-point custom made scale	Without assistance (human) Custom scale score 1, 2	Three-months
Ween et al. (37)	n=unknown/376 Age (years)=73 (SD 12) Gender=169 M, 207 F	Stroke, no SAH	Rehab	FIM	With assistance (human) FIM 1–5	19	FIM	Without assistance (human) FIM 6, 7	D/C (two-months)

Data recorded in the table, including number of nonambulatory participants, the outcome measure (in the case of multiple measures), and the time of follow-up measures were used for the meta-analyses. n, number of nonambulatory participants/number of participants in the sample; Acute, acute unit; rehab, rehabilitation unit; FIM, functional independence measure; MAS, Motor Assessment Scale for Stroke; BI, Barthel Index.

Quality

Of the 26 studies included in this review, 21 were of good methodological quality, and of the 17 entered into the meta-analyses, 13 were of good methodological quality. The sample was representative of the population in 62% of studies and partly representative in 35%, loss to follow-up was not associated with key characteristics in 88% of studies, outcome measurement was adequate in 96% of studies and important potential confounders were at least partly accounted for in 80% of studies.

Participants

The mean ages of participants across the studies ranged from 58 to 80 years old and the mean time of initial measures of walking ranged from two- to 31 days. Patients were managed either in a rehabilitation unit (n=15) or an acute unit (n=11).

Outcome measures

Although many different measurement tools were used to report walking, independent walking was reported in 96% of studies as the ability to walk alone or with an aid, but without human assistance. Nonambulatory was defined as FIM mobility and locomotion subscale score <6 (n=5), Barthel index mobility subscale score of 0, 5 or 10 (n=6), Barthel index mobility subscale score of 0–2 (n=3), MAS item 5 score 0–2 (n=2), COVs mobility scale item 5 score 0–4 (n=1), modified Rankin scale >3 (n=1), Hemiplegic Stroke Severity Scale >3 (n=1) and walking speed <0·15 m/s (n=1). The remaining studies (n=6) used scales developed by the authors to define nonambulatory. Independent walking data were available to be used in a meta-analysis from patients managed in a rehabilitation unit at three-months (n=9) and at six-months (n=3) and from those managed in an acute unit at three-months (n=6) and at six-months (n=3).

Probability of regaining independent walking

Rehabilitation unit

At three-months, the probability of initially nonambulatory stroke survivors regaining independent walking was examined by pooling data from nine studies (14, 18, 21, 25, 26, 28, 29, 32, 35) comprising 1373 participants (Fig. 3). The probability of independent walking in initially nonambulatory stroke patients was 0·60 (95% CI 0·47–0·72). Data were unavailable from three studies.

Fig. 3

Probability (95% CI) of patients after stroke who are initially nonambulatory regaining independent walking at three-months when managed in a rehabilitation unit by pooling data from nine studies (n=1373).

At six-months, the probability of initially nonambulatory stroke survivors regaining independent walking was examined by pooling data from three studies (18, 29, 30) comprising 444 participants (Fig. 4). The probability of independent walking had increased to 0·65 (95% CI 0·53–0·77).

Fig. 4

Probability (95% CI) of patients after stroke who are initially nonambulatory regaining independent walking at six-months when managed in a rehabilitation unit by pooling data from three studies (n=444).

At 12 months, the probability of initially nonambulatory stroke survivors regaining independent walking was examined by using data from one study (35) comprising 24 participants. The probability of independent walking had increased to 0·91 (95% CI 0·81–1·00). Data were unavailable from one study.

Acute unit

At three-months, the probability of initially nonambulatory stroke survivors regaining independent walking was examined by pooling data from five studies (2, 19, 20, 36, 38) comprising 634 participants (Fig. 5). The probability of independent walking in initially nonambulatory stroke patients was 0·39 (95% CI 0·27–0·52). Data were unavailable from four studies.

Fig. 5

Probability (95% CI) of patients after stroke who are initially nonambulatory regaining independent walking at three-months when managed in an acute unit by pooling data from five studies (n=634).

At six-months, the probability of initially nonambulatory stroke survivors regaining independent walking was examined by pooling data from two studies (19, 31) comprising 405 participants (Fig. 6). The probability of independent walking had increased to 0·69 (95% CI 0·46–0·92). Data were unavailable from two studies.

Fig. 6

Probability (95% CI) of patients after stroke who are initially nonambulatory regaining independent walking at six-months when managed in an acute unit by pooling data from two studies (n=405).

At 12 months, the probability of initially nonambulatory stroke survivors regaining independent walking was examined by using data from one study (24), comprising 34 participants. The probability of independent walking had increased to 0·74 (95% CI 0·59–0·88).

Discussion

This is the first systematic review to pool data from prospective, consecutive studies of patients who are nonambulatory in the first month after stroke to determine the probability of regaining independent walking. A meta-analysis of available data found that 60% of initially nonambulatory stroke patients managed in a rehabilitation unit regained independent walking at three-months and this increased to 65% by six-months after stroke. This differed from the findings for patients managed in an acute unit – at three-months only 39% of initially nonambulatory stroke patients in an acute unit regain independent walking. However, at six-months after stroke, a similar proportion of patients managed in an acute unit (69%) regained independent walking to those managed in a rehabilitation unit (Fig. 7).

Fig. 7

The probability of patients after stroke who are initially nonambulatory regaining independent walking when managed in a rehabilitation unit (closed circles) and an acute unit (open circles).

The overall quality of the included studies was good. The age and gender of participants was similar across the studies, as were the inclusion criteria to enter the studies. There were 15 studies that included participants managed in a rehabilitation setting, and 11 that included participants managed in an acute setting, which included both general acute units and designated stroke units. Initial measures of walking occurred earlier on average in acute units (six-days) than in rehabilitation units (19 days). A variety of measures of walking were used across the included studies, and not all studies defined independent walking. However, of the studies that did, all but one study defined independent walking as the ability to walk without human assistance. There were sufficient data for a meta-analysis to be included at three- and six-months for rehabilitation units and acute units. The number of participants included was 1373 in the rehabilitation unit meta-analysis and 634 in the acute unit meta-analysis at three-months. The good quality and similarity of the studies, as well as the number of participants available for the three-month meta-analyses, suggests that the results for the probability of regaining independent walking at three-months after stroke in patients who are initially nonambulatory are credible. Although there were data available for analysis at six- and 12 months in both settings, the number of participants is too small to draw valid conclusions.

A substantially greater proportion of patients managed in a rehabilitation unit regained independent walking at three-months compared with those managed in an acute unit. There are several possible explanations for this outcome. Patients admitted to a rehabilitation unit have generally been selected as being likely to benefit from a prolonged period of rehabilitation, thus increasing the probability of this population regaining independent walking, compared with an unselected population of patients managed in an acute unit after stroke. Patients admitted to an acute unit after stroke can include those with severe stroke, including those with significant physical and cognitive impairment. The severity of impairments in these patients would also contribute to a reduced probability of regaining independent walking for patients managed in an acute unit. Moreover, the outcome of stroke is associated with intensity of therapy (39) and rehabilitation units are designed and resourced to provide a greater intensity of therapy than acute units, thus increasing the probability of improved walking outcomes in these patients.

There was variability in the number of days after stroke that the initial measure of walking was completed across studies. The correlation between the average day the participants were measured as nonambulatory and the probability of walking independently at three-months was high in both a rehabilitation unit (r=0·88; P=0·002) and an acute unit (r=0·70; P=0·19). This is not surprising given there is the potential for substantial recovery in walking ability in the first month after stroke (40), and a patient who is still nonambulatory 30 days after stroke is likely to have had a devastating stroke, which is not necessarily the case in patients who are nonambulatory at only three-days after stroke. Therefore, we carried out a post hoc regression analysis taking into account the number of days after stroke that measurement of nonambulatory status occurred. The prediction equation for determining the probability of independent walking at three-months in patients who are nonambulatory in the first month after stroke in a rehabilitation unit is:

probability of walking=1·24–(0·034 × day after stroke measured as nonambulatory)

For an acute unit the prediction equation is:

probability of walking=0·66–(0·04 × day after stroke measured as nonambulatory)

For example, in a rehabilitation unit a patient who is measured as nonambulatory 14 days after stroke has a 76% chance of walking independently at three-months [0·76=1·24–(0·034 × 14)], compared with a patient in an acute unit who is measured as nonambulatory 14 days after stroke who has a 10% chance of walking independently at three-months [0·1=0·66–(0·04 × 14)]. These equations can be used to refine the prediction of probability of individual patients regaining independent walking at three-months after stroke. This more accurate information could be utilized to allocate resources amongst individual patients, as well as to facilitate decisions about discharge destination. It also provides clinical services with a benchmark for evaluating the effectiveness of service provision for patients who are initially nonambulatory after stroke and can also be used as a guide to educate patients and carers of the probability of regaining independent walking. Moreover, these equations can be utilized in clinical research as a benchmark for the performance of control groups in trials examining walking outcome after stroke. For instance in two recent randomized trials examining the effectiveness of treadmill training in nonambulatory participants (41, 42), the outcome of the control group in both studies was consistent with the proportion of independent walkers predicted by the equation, suggesting that the control groups in both studies achieved a reasonable outcome in terms of walking independence.

There are several limitations of this review. Firstly, data were unable to be extracted from nine studies, thus reducing the power of the results. There were substantially more data available for the meta-analyses of walking outcome at three-months than at any other time point, and only two studies had data available for analysis at 12 months after stroke, suggesting that the findings at six- and 12 months in both settings are less robust than the findings at three-months after stroke. Secondly, in line with best practice (43), inclusion in this systematic review required studies to be consecutive with inception cohorts, consequently randomized trials were not screened. However, it has been suggested that the control groups in randomized trials may provide ‘hidden’ information about probability of outcomes (43). As a result, some potentially relevant data may have been excluded from the analysis.

In conclusion, this systematic review found that in a rehabilitation unit, the probability of regaining independent walking for patients who are nonambulatory in the first month after stroke is 60% at three-months. The probability of regaining walking is lower for patients managed in an acute unit at only 39% by three-months. The prediction may be refined by using the time in the first month after stroke that the patient remains nonambulatory, to predict walking outcome at three-months for individual patients in both rehabilitation and acute settings. This information may be used clinically to make decisions about allocation of rehabilitation resources, education of patients and carers, and for discharge planning.

Footnotes

Acknowledgements

We would like to thank Dr Liesbet De Wit, Ms Elisabeth Ekstrand, Dr Cesar Minelli, Dr John Reid, Dr Disa Sommerfeld and Dr Jon Ween for providing additional information for the meta-analyses. We would also like to thank Ms Irmina Nahon, Mrs Helen Preston and Mr Nicholas Stanton for their time and skill in translating papers into English.

References

1 Solomon

Glick

Russo

Lee

Schulman

. Patient preferences for stroke outcomes. Stroke 1994; 25:1721–5.

2 Sommerfeld

von Arbin

. Disability test 10 days after acute stroke to predict early discharge home in patients 65 years and older. Clin Rehabil 2001; 15:528–34.

3 Van de Port

Kwakkel

Schepers

VPM

Lindeman

. Predicting mobility outcome one year after stroke: a prospective cohort study. J Rehabil Med 2006; 38:218–23.

4 Kwakkel

Wagenaar

Kollen

Lankhorst

. Predicting disability in stroke – a critical review of the literature. Age Ageing 1996; 25:479–86.

5 Meijer

Ihnenfeldt

van Limbeek

Vermeulen

de Haan

. Prognostic factors in the subacute phase after stroke for the future residence after six months to one year. A systematic review of the literature. Clin Rehabil 2003; 17:512–20.

6 Koton

Tanne

Green

Bornstein

. Mortality and predictors of death 1 month and 3 years after first-ever ischemic stroke: data from the first national acute stroke Israeli survey. Neuroepidemiology 2010; 34:90–6.

7 Payne

Huybrechts

Caro

Craig Green

Klittich

. Long term cost-of-illness in stroke: an international review. Pharmacoeconomics 2002; 20:813–25.

8 Pound

Gompertz

Ebrahim

. A patient-centred study of the consequences of stroke. Clin Rehabil 1998; 12:338–47.

9 Yashin

Akushevich

Ukraintseva

Akushevich

Arbeev

Kulminski

. Trends in survival and recovery from stroke: evidence from the National Long-Term Care Survey/Medicare Data. Stroke 2010; 41:563–5.

10.

10 Hayden

Côté

Bombardier

. Evaluation of the quality of prognosis studies in systematic reviews. Ann Intern Med 2006; 144:427–37.

11.

11 Higgins

JPT

Green

. (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.2 [updated September 2009]. The Cochrane Collaboration, 2009. Available at http://www.cochrane-handbook.org (Accessed 10 January 2011)

12.

12 Bax

Ikeda

Tsuruta

Moons

. Development and validation of MIX: comprehensive free software for meta-analysis of causal research data. BMC Med Res Methodol 2006; 13:50.

13.

13 Bax

Ikeda

Fukui

Yaju

Tsuruta

Moons

SGM

. More than numbers: the power of graphs in meta-analysis. Am J Epidemiol 2008; 169:249–55.

14.

14 Bacciglieri

Basaglia

Bortone

Salvadori

Bertocchi

Starita

. Prognostic evaluation of walking ability in hemiplegic patients using an objective measurement scale. Eur Medicophys 1995; 31:1–11.

15.

15 Baer

Smith

. The recovery of walking ability and subclassification of stroke. Physiother Res Int 2001; 6:135–44.

16.

16 Bagg

Pombo

Hopman

. Effect of age on functional outcomes after stroke rehabilitation. Stroke 2002; 33:179–85.

17.

17 Daviet

Verdie-Kessler

Stuit

. Early prediction of functional outcome one year after initial unilateral hemispheric stroke. Ann Readaptation Med Phys 2006; 49:49–56.

18.

18 De Wit

Putman

Schuback

. Motor and functional recovery after stroke: a comparison of 4 European rehabilitation centers. Stroke 2007; 38:2101–7.

19.

19 Friedman

. Gait recovery after hemiplegic stroke. Int Disabil Studies 1991; 12:119–22.

20.

20 Jorgensen

Nakayama

Raaschou

Olsen

. Recovery of walking function in stroke patients – The Copenhagen Stroke study. Arch Phys Med Rehabil 1995; 76:27–32.

21.

21 Kuys

Bew

Lynch

Morrison

Brauer

. Measures of activity limitation on admission to rehabilitation after stroke predict walking speed at discharge: an observational study. Aust J Physio 2009; 55:265–8.

22.

22 Loewen

Anderson

. Predictors of stroke outcome using objective measurement scales. Stroke 1990; 21:78–81.

23.

23 Macciocchi

Diamond

Alves

Mertz

. Ischemic stroke: relation of age, lesion location, and initial neurologic deficit to functional outcome. Arch Phys Med Rehabil 1998; 79:1255–7.

24.

24 Minelli

Fen

Minelli

DPC

. Stroke incidence, prognosis, 30-day, and 1-year case fatality rates in Matao, Brazil – A population-based prospective study. Stroke 2007; 38:2906–11.

25.

25 Olsen

. Arm and leg paresis as outcome predictors in stroke rehabilitation. Stroke 1990; 21:47–251.

26.

26 Paolucci

Bragoni

Coiro

. Quantification of the probability of reaching mobility independence at discharge from a rehabilitation hospital in nonwalking early ischemic stroke patients: a multivariate study. Cerebrovasc Dis 2008; 26:16–22.

27.

27 Patel

Duncan

Lai

Studenski

. The relation between impairments and functional outcomes poststroke. Arch Phys Med Rehabil 2000; 81:1357–63.

28.

28 Petrilli

Durufle

Nicolas

Pinel

Kerdoncuff

Gallien

. Prognostic factors in the recovery of the ability to walk after stroke. Cerebrovasc Dis 2002; 11:330–5.

29.

29 Piron

Piccione

Tonin

Dam

. Clinical correlation between motor evoked potentials and gait recovery in poststroke patients. Arch Phys Med Rehabil 2005; 86:1874–8.

30.

30 Prevo

AJH

Dijkman

MMTT

Le Fèvre

FAM

. Impairment and disability in patients with a severe ischaemic cerebral infarction at admission to the rehabilitation centre and six months after the stroke. Ned Tijdschr Geneeskd 1998; 142:637–40.

31.

31 Reid

Gubitz

Dai

. Predicting functional outcome after stroke by modelling baseline clinical and CT variables. Age Ageing 2010; 39:360–6.

32.

32 Singh

Hunter

Philip

Todd

. Predicting those who will walk after rehabilitation in a specialist stroke unit. Clin Rehabil 2006; 20:149–52.

33.

33 Thommessen

Bautz-Holter

Laake

. Predictors of outcome of rehabilitation of elderly stroke patients in a geriatric ward. Clin Rehabil 1999; 13:123–8.

34.

34 Van Bennekom

Jelles

Lankhorst

Bouter

. The rehabilitation activities profile – a validation-study of its use as a disability index with stroke patients. Arch Phys Med Rehabil 1995; 76:501–7.

35.

35 Viosca

Lafuente

Martinez

Almagro

Gracia

Gonzalez

. Walking recovery after an acute stroke: assessment with a new functional classification and the Barthel index. Arch Phys Med Rehabil 2005; 86:1239–44.

36.

36 Wade

Wood

Hewer

. Recovery after stroke – the first 3 months. J Neurol Neurosurg Psychiatry 1985; 48:7–13.

37.

37 Ween

Alexander

D'Esposito

Roberts

. Factors predictive of stroke outcome in a rehabilitation setting. Neurology 1996; 47:388–92.

38.

38 Ekstrand

Ringsberg

Pessah-Rasmussen

. The physiotherapy clinical outcome variables scale predicts length of hospital stay, discharge destination and future home facility in the acute comprehensive stroke unit. J Rehabil Med 2008; 40:524–8.

39.

39 Kwakkel

van Peppen

Wagenaar

. Effects of augmented exercise therapy time after stroke: a meta-analysis. Stroke 2004; 35:2529–36.

40.

40 Kollen

van de Port

Lindeman

Twisk

Kwakkel

. Predicting improvement in gait after stroke – a longitudinal prospective study. Stroke 2005; 36:2676–80.

41.

41 Ada

Dean

Morris

Simpson

Katrak

. Randomised trial of treadmill walking with body weight support to establish walking in subacute stroke: the MOBILISE trial. Stroke 2010; 41:1237–42.

42.

42 Pohl

Kreischa

Werner

. Repetitive locomotor training and physiotherapy improve walking and basic activities of daily living after stroke: a single-blind, randomized multicentre trial (DEutsche GAngtrainerStudie, DEGAS). Clin Rehabil 2007; 21:17–27.

43.

43 Moons

KGM

Royston

Vergouwe

Grobbee

Altman

. Prognosis and prognostic research: what, why, and how? BMJ 2009; 338:b375.