http://informahealthcare.com/dre ISSN 0963-8288 print/ISSN 1464-5165 online Disabil Rehabil, 2015; 37(11): 942–950 ! 2014 Informa UK Ltd. DOI: 10.3109/09638288.2014.948137

RESEARCH PAPER

Predicting sickness impact profile at six months after stroke: further results from the European multi-center CERISE study Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

C. Stummer1, G. Verheyden1*, K. Putman2, W. Jenni3, W. Schupp4, and L. De Wit2 1

Department of Rehabilitation Sciences, KU Leuven, Leuven, Belgium, 2Department of Medical Sociology, Faculty of Medicine and Pharmacy, Vrije Universiteit Brussel, Brussel, Belgium, 3Rehaklinik Rheinfelden, Rheinfelden, Switzerland, and 4Department of Neurologie/Neuropsychologie, Fachklinik, Herzogenaurach, Germany Abstract

Keywords

Purpose: To develop prognostic models and equations for predicting participation at six months after stroke. Methods: This European prospective cohort study recruited 532 consecutive patients from four rehabilitation centers. Participation was assessed at six months after stroke with the Sickness Impact Profile (SIP), which consists of a physical, psychosocial and independent dimension. Twenty-six independent variables on admission to the rehabilitation center and 13 additional variables measured at two months post stroke were included in the analysis. A multiple logistic regression analysis was conducted predicting good participation (SIP520%). Sensitivity, specificity, positive and negative predictive values were calculated. Results: The prognostic models for the three dimensions provided independent predictors containing demographics, complications, diagnostic, and disability measures. Sensitivity ranged from 64–84%, specificity 66–85%, positive predictive value 70–78%, and negative predictive value 76–87%. Barthel Index on admission, Euroqol Health State at two months and Caregiver Strain Index at two months were retained in all prediction models. Conclusions: A combination of variables was found in the prognostic models of the three dimensions of the SIP at six months after stroke. Already from the early beginning of stroke rehabilitation it seems important to focus on personal activities of daily living as well as caregivers’ strain.

Barthel index, prediction, prognosis, sickness impact profile, stroke History Received 27 October 2013 Revised 17 July 2014 Accepted 21 July 2014 Published online 29 August 2014

ä Implications for Rehabilitation 





Prognostic factors predicting participation, measured by the three dimensions of the Sickness Impact Profile at six months post stroke include demographic variables, post-stroke complications, diagnostic parameters and disability measures. Significant prognostic variables for all three dimensions of the Sickness Impact Profile were a higher Barthel Index score on admission to the rehabilitation center, a higher Euroqol Health State score at two months post stroke and a lower Caregiver Strain Index score at two months post stroke. Early stroke therapy should therefore further emphasize rehabilitation of personal activities of daily living such as mobility, walking, feeding, dressing, and toilet use, as well as considering strategies to reduce caregiver strain such as giving support, providing information and training carers.

Introduction When a person after stroke is admitted to a rehabilitation center, it is important to inform patient, family and carers about therapy planning and the achievable level of functional recovery [1,2]. Predicting outcome is part of optimized stroke management in order to set relevant and realistic therapy goals [2].

*Joint first author Address for correspondence: Dr. Liesbet de Wit, Department of Medical Sociology, Faculty of Medicine and Pharmacy, Vrije Universiteit Brussel, Laarbeeklaan 103, 1090 Brussel, Belgium. Tel: +32-2-4774767. E-mail: [email protected]

According to the domains of the International Classification of Functioning, Disability and Health (ICF), there are different levels of predicting rehabilitation outcome post stroke [3]. Most studies focus on predicting the activity level, as demonstrated by a recent systematic review of Veerbeek et al. [2], which included 48 studies. They reported that the most frequently retained prognostic factors of outcome in activities of daily living (ADL) of 6 high-quality studies [4–9] were age [4–6,8,9], and National Institute of Health Stroke Scale score (NIHSS) on admission to the acute hospital [5,7–9]. The above-mentioned studies reported prognostic factors, but did not present an applicable prediction equation. Prediction equations enable the multidisciplinary team to provide individual stroke predictions. Literature shows that there is only one study [10] that provides a prediction equation to

Predicting participation after stroke

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

DOI: 10.3109/09638288.2014.948137

predict outcome at six months post stroke. This model, however, predicted ambulation and ADL independence which does not clearly represent the domain of participation of the ICF. Previous prediction studies were not primarily designed to investigate participation, but rather to assess patients on their activity level. A recent systematic review [11] of instruments assessing participation pointed out that there is an increased focus on the assessment of a person’s participation. With the aim of predicting participation, it is necessary to have a clear definition of participation. In literature, various definitions can be found [11], but the ICF [3] defines participation as: ‘‘A person’s involvement in a life situation. It represents the societal perspective of functioning’’. Whiteneck et al. [12] stated that ‘‘activities are tasks performed by individuals while participation is the social role performance as a member of society with or for others’’. The current study will use the definition of the ICF, to describe participation, which we believe is in line with the definition used by the above authors [12]. Two studies [13,14] reported that physical recovery is to a great extent achieved at six months after stroke. Duncan et al. [14] also highlighted that especially participation should not be measured sooner than at six months post stroke, as patients need this time to stabilize their social situation. Mayo et al. [15] reported that 54% of their population-based sample experienced limitations in higher-level activities of daily living, such as housework and shopping, and a total of 65% experienced restrictions in reintegration into community activities at six months post stroke. The results of van der Zee et al. [16] also showed that almost one out of two stroke survivors experienced restrictions in physical exercise, household tasks and outdoor activities one year post stroke. It may be concluded that participation is one of the most relevant, but neglected topics in stroke rehabilitation. This emphasizes the importance of investigating prognostic models and equations to predict participation at six months after stroke. Until now, there is no study that included different domains of participation (such as learning and applying knowledge, general tasks and demands, communication, mobility, self-care, domestic life, interpersonal interactions and relationships, major life areas, community, social and civic life) when predicting outcome at six months after stroke. The CERISE study (Collaborative Evaluation of Rehabilitation in Stroke across Europe), a European, multicenter, prospective cohort study provides the opportunity to develop prognostic models for predicting participation. CERISE included the Sickness Impact Profile, which measures six ICF participation domains (communication, mobility, self-care, domestic life, interpersonal interactions and relationships, and major life areas) at six months after stroke. Therefore, the aim of this study was to develop prediction models, as well as prediction equations for each of the three dimensions of the Sickness Impact Profile (SIP); the physical (ambulation, mobility and body care and movement domains), psychosocial (social interaction, alertness behavior, emotional behavior and communication domains) and independent dimension (sleep and rest, eating, work, home management and recreation and pastimes domains). Prediction models are derived for the SIP at six months post stroke, based on variables collected on admission to the rehabilitation center and at two months post stroke.

Methods

943

Nottingham University Hospitals, Nottingham, United Kingdom; RehaClinic, Zurzach, Switzerland; and Fachklinik, Herzogenaurach, Germany. Patients’ inclusion criteria were: (1) first-ever stroke as defined by the World Health Organization (WHO) [18]; (2) score on the Rivermead Motor Assessment [19] – Gross Motor Function (RMA-GF)  11 out of 13, and/or score on the Rivermead Motor Assessment Leg/Trunk (RMA-LT)  8 out of 10 and/or score on the Rivermead Motor Assessment Arm (RMA-A)  12 out of 15 on admission to the center; and (3) age between 40 and 85 years. The exclusion criteria were: (1) other neurological impairments with permanent damage; (2) stroke-like symptoms caused by subdural hematoma, tumor, encephalitis, or trauma, (3) admission to the center 46 weeks after stroke (to exclude chronic stroke patients); (4) no informed consent; and (5) pre-stroke Barthel Index (BI) [20] 550 out of 100. The CERISE study was approved by the ethics committee of each center. Dependent (outcome) variables The level of participation was assessed at six months after stroke with the Sickness Impact Profile (SIP) [21]. The SIP was sent out via postal mail to the patients for self-administration before a standardized home assessment at six months after stroke was conducted. At that time, the researcher verified the completeness of the SIP and completed the SIP together with the patient if necessary. The SIP is a 136-item generic questionnaire and is a behavioral-based measure of perceived health status [21]. It contains 12 subscales that can be categorized in three dimensions. The independent dimension consist of five subscales (sleep and rest, eating, work, home management, recreation and pastimes), the physical dimension of three subscales (ambulation, mobility, body care and movement), and the psychosocial dimension of four subscales (social interaction, alertness behavior, emotional behavior and communication) [21]. The questionnaire takes about 30 minutes to complete [22]. The total score for each dimension ranges between 0 and 100%, with 0% representing no restriction and a score 420% representing severe restriction in participation [23]. Test-retest reliability of the SIP was found to be high (ICC ¼ 0.88–0.92) [21], as well as the Cronbach’s alpha coefficient (0.94) [21]. External validity and clinical applicability have been determined [21]. Reliability does not change with different administration circumstances (self- or interviewadministered) [21,24]. Independent variables A total of 26 potentially prognostic variables, comprising demographics, secondary complications, diagnostic and disability measures, and socio-economic factors were recorded on admission to the rehabilitation center. Additionally, nine variables measured at two months after stroke, as well as four subtracted variables (two months variable subtracted by admission variable), encompassing disability measures, quality of life and caregivers’ strain, were calculated. The four subtracted variables were: Rivermead assessment-Gross motor function (RMA-GF S), Rivermead assessment-leg and trunk (RMA-LT S), Rivermead assessment-arm (RMA-A S) and Barthel Index (BI S). Because literature on prognosis of participation after stroke is limited, variables were selected based on general stroke prediction literature. The list of independent variables is shown in Table 1.

Subjects and settings Participants of this study were enrolled in the CERISE study [17], which compared stroke rehabilitation between four European rehabilitation centers: University Hospital, Leuven, Belgium;

Data analysis Patient characteristics were described with frequencies, means and standard deviations or medians, interquartile ranges and

944

C. Stummer et al.

Disabil Rehabil, 2015; 37(11): 942–950

Table 1. Patient characteristics.

Table 1. Continued

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

Variable

Result

Assessed on admission to the center Center: n (%) Pellenberg (BE) Nottingham (UK) Zurzach (CH) Herzogenaurach (DE)

127 135 135 135

Demographics Male gender: n (%) Age: mean (SD)

283 (53.2) 69.4 (10.3)

Monthly income: Low n (%) Moderate n (%) High n (%) Unknown n (%)

98 261 113 60

Education: Low n (%) High n (%) Unknown n (%)

398 (74.8) 110 (20.7) 24 (4.5)

Pathology Type of stroke: n (%) Haemorrhagic insult Ischemic infarct Unknown

77 (14.0) 445 (84.0) 10 (2.0)

Side of lesion: n (%) Left Right Both Unknown

219 261 47 5

(41.2) (49.1) (8.8) (0.9)

111 354 219 253 104 97

(20.9) (66.6) (41.2) (47.6) (19.5) (18.2)

25 135 426 68 106

(4.7) (25.4) (80.1) (12.8) (19.9)

Clinical scores Pre-Stroke Rankin: n (%) No symptoms No significant disability Slight disability Moderate disability Moderately severe disability TSOA (days): median (IQR, min–max) Pre stroke BI: median (IQR, min–max) NIHSS: median (IQR, min–max) RMA-GF: median (IQR, min–max) RMA-LT: median (IQR, min–max) RMA-A: median (IQR, min–max) BI: median (IQR, min–max)

393 84 22 32 1 19 100 6 5 6 4 55

(73.9) (15.8) (4.1) (6) (0.2) (14–27, 4–42) (100–100, 55–100) (3–10, 0–23) (1–9, 0–13) (2–8, 0–10) (1–11, 0–15) (30–80, 0–100)

Assessed at two months after stroke RMA-GF 2: median (IQR, min–max) RMA-LT 2: median (IQR, min–max) RMA-A 2: median (IQR, min–max) BI 2: median (IQR, min–max) EADL 2: median (IQR, min–max) EQHS 2: median (IQR, min–max) EQ-VAS 2: median (IQR, min–max) HADS 2: median (IQR, min–max) CSI 2: median (IQR, min–max) RMA-GF S: median (IQR, min–max)

9 8 9 80 5 62.4 60 4 2 2

(3–11, 0–13) (3–9, 0–10) (1–13, 0–15) (50–98.7, 0–100) (2–12, 0–22) (48.3–77.9, 15.5–100) (50–75, 0–100) (2–8, 0–20) (0–5, 0–13) (0–4, 13–11)

Risk Factors Diabetes Mellitus: n (%) History of high blood pressure: n (%) Hyperlipidaemia: n (%) Ever smoked: n (%) Atrial fibrillation: n (%) Complications & Comorbidity Secondary complications: n (%) Neurosurgical intervention: n (%) Chronic heart disease: n (%) Urinary incontinence: n (%) Myocardial infarction: n (%) Swallowing problem: n (%)

Variable RMA-LT S: median (IQR, min–max) RMA-A S: median (IQR, min–max) BI S: median (IQR, min–max)

(23.9) (25.4) (25.4) (25.4)

Result 1 (0–2, 10–10) 1 (0–3, 13–15) 15 (5–25, 95–75)

n (%), number of positives (percentage); BI, Barthel Index (range: 0– 100); IQR, Interquartile range (percentile 25–percentile 75); NIHSS, National Institutes of Health Stroke Scale (range: 0–23); RMA-GF, Rivermead Motor assessment-Gross motor function (range: 0–13); RMA-LT, Rivermead Motor assessment-leg and trunk (range: 0–10); RMA-A, Rivermead Motor assessment-arm (range: 0–15); pre stroke BI pre stroke Barthel Index (range: 55–100); Monthly income: 1 ¼ low income (560% of median national equivalent income), 2 ¼ moderate (60–120%), 3 ¼ high (4120%); Education: 1 ¼ low education (below or equal to secondary level), 2 ¼ high education (upper secondary level or higher); TSOA, time between stroke onset and admission to the stroke rehabilitation unit in days; SD, Standard Deviation; 2, assessed at two months after stroke ; EADL, Extended Activities of Daily Living (range: 0–22); EQHS, Euroqol Health State; EQVAS, Euroqol Visual Analogue Scale; HADS, Hospital Anxiety and Depression Scale (range: 0–42); CSI, Caregiver Strain Index (range: 0–13); S, two months variable minus admission variable; N range: 455–532.

(18.4) (49.1) (21.2) (11.3)

(continued )

minimum-maximums, depending on the type of data (Table 1). At six months after stroke, the total scores for each SIP dimension were dichotomized in scores between 0 and 20% (no to low participation restriction) and scores 420% (severe participation restriction). A univariate binary logistic regression was performed to identify significant univariate predictors for each of the three SIP dimensions. Variables with a p value50.1 were included in the subsequent multiple binary logistic regression analysis. Spearman rank correlation coefficients were calculated to check for multicollinearity. Correlation coefficients 40.80 were considered being collinear. A multiple binary logistic regression analysis was performed to generate prediction models and equations for each of the three SIP dimensions. All independent variables with p50.1 in the univariate analysis were entered forward stepwise into the models. The probability for entry of a variable was set at p50.05 and for removal at p40.1. Level of significance for retaining variables in the multiple logistic regression model was set at p50.05. Two-by-two tables were created to display actual and predicted outcome. Accuracy, sensitivity, specificity, negative and positive predictive value of the prediction equations were calculated based on these tables. Data were analyzed with IBM SPSS Statistics version 19 (Chicago, IL).

Results Participants A total of 532 consecutive patients with stroke were recruited from the multidisciplinary rehabilitation units. There were 135 patients admitted in all centers, except in Belgium (n ¼ 127). Patients’ characteristics are shown in Table 1. At six months post stroke, 77 participants were lost-to-follow-up: 18 died, 54 refused to participate, and 5 could not be assessed (missed assessment or poor medical condition). Univariate logistic regression analysis The results of the univariate logistic regression analysis of the physical, psychosocial and independent dimension of the SIP are shown in Tables A, B and C in the Appendix. A total of 31

Predicting participation after stroke

DOI: 10.3109/09638288.2014.948137

variables for the physical, 28 variables for the psychosocial, and 31 variables for the independent dimension showed significant results (p50.1). Nine variables showed multicollinearity: RMA-GF, RMA-LT, RMA-A, BI all assessed on admission; and RMA-GF 2, RMA-LT 2, RMA-A 2, BI 2, Extended Activities of Daily Living (EADL 2) all assessed at two months after stroke. Out of these nine variables, Barthel Index-admission score (BI), Barthel Index-two months score (BI-2) and Rivermead Motor Assessment-Armadmission score (RMA-A) were not multicollinear with each other. Consequently, these three variables were entered into the multiple models.

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

Multiple logistic regression analysis Tables 2–4 show the results of the multiple regression analyses for the three SIP dimensions as well as the prediction equations. The model predicting good outcome (score520%) for the physical SIP dimension (Table 2) included center, hemorrhagic stroke, younger age, higher Barthel Index score on admission, higher Euroqol Health State score at two months, lower Caregiver Strain Index score at two months, and greater improvement for RMA-GF score from admission to two months. Sensitivity was 81%, specificity 85%, positive predictive value 78%, and negative predictive value 87%. The model predicting good outcome (score520%) for the psychosocial SIP dimension (Table 3) included side of lesion, Table 2. Multiple regression model and equation for the SIP physical dimension (good outcome 520%). Variable

Exp (B)

95% CI for Exp (B)

p Value

1.37 2.56 0.95 1.05 1.05 0.88 1.30

1.03–1.82 1.19–5.56 0.93–0.98 1.03–1.06 1.03–1.08 0.79–0.96 1.14–1.49

0.03 0.02 0.00 0.00 0.00 0.01 0.00

Center Type of stroke Age BI EQ-HS 2 CSI 2 RMA-GF S P/(1-P) ¼ e5.31  0.32

(Type

lower pre stroke Rankin scale score, having no swallowing problems, shorter time between stroke onset and admission to the rehabilitation center, higher Barthel Index score on admission, higher Euroqol Health State score at two months, lower Caregiver Strain Index score at two months, and lower Hospital Anxiety and Depression Scale score at two months. Sensitivity was 84%, specificity 66%, positive predictive value 77%, and negative predictive value 76%. The model predicting good outcome (520%) for the independent SIP dimension (Table 4) included having hyperlipidaemia, higher Barthel Index score on admission, higher Euroqol Health State score at two months, and lower Caregiver Strain Index score at two months. Sensitivity was 64%, specificity 84%, positive predictive value 70%, and negative predictive value 80%. Post-hoc analysis Table 5 presents the post-hoc analysis for center as it was retained in the model predicting participation restriction in the physical dimension of the SIP. Results showed that patients admitted to the Swiss and German centers had a significantly higher chance of having a good outcome for the physical dimension (520%) at six months after stroke in comparison with patients admitted to the Belgian and British center, after correction for the other variables retained in the multiple model. The post-hoc results for ‘‘side of lesion’’ in the psychosocial dimension showed no significant differences between left, right or both-sided lesion (Table 6).

Discussion In this study, we presented three prognostic models predicting participation at six months after stroke, based on the three dimensions (physical, psychosocial and independent) of the Sickness Impact Profile. In each model, a combination of variables was retained. The following diagnostic measures were retained: type of stroke (physical dimension), side of lesion (psychosocial dimension), and hyperlipidaemia (independent

1 ¼ Pellenberg/2 ¼ Nottingham/3 ¼ Zurzach/

(Center,

4 ¼ Herzogenaurach)  0.94

945

of

stroke,

ischemic ¼ 0/hemorrhagic ¼ 1) + 0.05

(Age)  0.05 (BI)  0.06 (EQ-HS 2) + 0.13 (CSI 2) – 0.27 (RMA-GF-S)

BI, Barthel Index; EQ-HS, Euroqol Health State; CSI, Caregiver Strain Index, 2, assessed at two months after stroke; RMA-GF, Rivermead assessment-Gross motor function; S, two months variable minus admission variable. Table 3. Multiple regression model and equation for the SIP psychosocial dimension (good outcome 520%).

Table 4. Multiple regression model and equation for the SIP independent dimension (good outcome 520%). Variable

Exp (B)

95% CI for Exp (B)

p Value

Hyperlipidaemia BI EQ-HS 2 CSI 2

1.64 1.03 1.04 0.91

1.01–2.63 1.02–1.04 1.03–1.06 0.84–0.98

0.04 0.00 0.00 0.01

P/(1  P) ¼ e5.16  0.49

(Hyperlipidaemia,

no ¼ 0/yes ¼ 1)  0.03

(BI)  0.05

(EQ-HS 2) + 0.09 (CSI 2)

Variable Side of lesion PSR Swallowing problems TSOA BI EQ-HS 2 CSI 2 HADS 2

Exp (B)

95% CI for Exp (B)

p Value

1.96 0.72 0.51 0.97 1.01 1.02 0.88 0.84

1.32–2.94 0.53–0.99 0.27–0.95 0.94–0.99 1.00–1.02 1.00–1.04 0.83–0.95 0.78–0.91

0.00 0.04 0.04 0.01 0.04 0.01 0.00 0.00

PSR, Pre-Stroke Rankin; TSOA, time between stroke onset and admission to the stroke rehabilitation unit in days; BI, Barthel Index; EQ-HS, Euroqol Health State; CSI, Caregiver Strain Index; HADS, Hospital Anxiety and Depression Scale; 2, assessed at two months after stroke. P/(1-P) ¼ e0.32  0.67 (Side of lesion, left ¼ 1/right ¼ 2/both ¼ 3) + 0.32 (PSR) + 0.68 (Swallowing

problems,

no ¼ 1/yes

2) + 0.12 (CSI 2) + 0.17 (HADS 2)

n ¼ 2) + 0.03

(TSOA)  0.01

(BI)  0.02

(EQ  HS

BI, Barthel Index; EQ-HS, Euroqol Health State; CSI, Caregiver Strain Index; 2, assessed at two months after stroke. Table 5. Post hoc analysis for center within SIP physical dimension model. Comparison BE versus UK BE versus CH BE versus DE UK versus CH UK versus DE CH versus DE

Exp (B)

95% CI for Exp (B)

Sig.

1.59 2.17 1.64 2.94 1.67 0.93

0.86–2.94 1.61–2.94 1.35–2.00 1.72–5.26 1.27–2.22 0.55–1.59

0.14 0.00 0.00 0.00 0.00 0.81

SIP indicates, Sickness Impact Profile; BE, Belgian center; UK, British center; CH, Swiss center; DE, German center; CI, confidence interval.

946

C. Stummer et al.

Disabil Rehabil, 2015; 37(11): 942–950

Table 6. Post hoc analysis for side of lesion within SIP psychosocial dimension model. Comparison Left versus right Left versus both Right versus both

Exp (B)

95% CI for Exp (B)

Sig.

1.37 1.30 1.22

0.92–2.04 0.90–1.85 0.61–2.50

0.13 0.16 0.57

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

SIP indicates Sickness Impact Profile; CI, confidence interval.

dimension). Demographics were retained in two models: center and age (physical dimension), and time between stroke onset and admission to the rehabilitation center (psychosocial dimension). Barthel Index on admission and Euroqol Health State and Caregiver Strain Index at two months were retained in all three models. Additionally, the improvement between admission and two months after stroke on the Rivermead Motor AssessmentGross motor function was retained in the physical dimension, and the pre-stroke Rankin scale score as well as the Hospital Anxiety and Depression Scale score at two months in the psychosocial dimension. Finally, swallowing problems was retained in the psychosocial dimension. None of the included socio-economic factors (education and income) were retained in any of the prediction models. SIP physical dimension Center was retained as one of the strongest predictors in the model, which limits the generalizability of this model. In essence, the physical dimension model can only be used in one of the four studied centers. The post-hoc analysis showed that patients admitted to the Swiss and German center had a higher chance of having a good level of participation at six months after stroke compared to patients admitted to the Belgian and British centers. These findings are in line with previous results presented by De Wit et al. [17]. Earlier results from the CERISE project [17,25,26] showed that stroke rehabilitation is different across Europe. Average therapy time ranged from 1 hour/day in the British center to 2 hours and 45 minutes/day in the Swiss center [25]. Time spent by therapists on direct patient care was the highest in the German center [26]. In the first six months after stroke, gross motor and functional recovery were also significantly better in the German and Swiss centers compared to the British center. Our post-hoc analysis of center showed similar results. This result can be explained by the difference in stroke rehabilitation management across the participating centers. It is accepted that more intensive rehabilitation results in better recovery [27,28], but would a better motor and functional recovery result in a higher level of participation? Our results should be generalized with caution as already suggested by Putman et al. [29], that the centers may not be representative of the individual countries [29]. Nevertheless, this model showed the highest sensitivity (81%) and specificity (85%), which implies the best explanatory power, compared to the other models. We identified another prognostic factor with a significant odds ratio (1.30) that can be influenced by physiotherapists as well as by the whole rehabilitation team: Gross motor function improvement between admission to the rehabilitation center and two months after stroke (RMA-GF S). A greater improvement was associated with a higher chance to have a good physical participation at six months after stroke. Therapists should be aware of this because this can be influenced by increasing the therapy time per patient per day [27,28], for example with individualized or group therapy. In times of working as cost effective as possible, group therapy such as task-oriented circuit

training might be an appropriate way of delivering therapy related to gross motor function. Furthermore, to increase the chance of reaching a good level of participation (physical dimension) at six months post stroke, it is important for the rehabilitation team to work on improvements in the Euroqol Health State and Caregiver Strain Index. Therefore, it might be necessary to assist caregivers, and to give emotional and practical support such as providing information on rehabilitation processes and methods for non-trained caregivers, as suggested by studies on functional and psychosocial outcome [30,31]. SIP psychosocial dimension The most important factor in the psychosocial dimension, which also can be influenced by therapists in the acute rehabilitation setting, are swallowing problems. If patients can improve swallowing and be in the group of having no swallowing problems on admission to the rehabilitation center, then patients have a significantly higher chance for a better psychosocial dimension of participation at six months. In line with our findings, dysphagia is known as an important factor for predicting functional outcome after stroke [32]. Four measures were retained in the model of the psychosocial dimension that could be influenced by the rehabilitation team: Barthel Index on admission, Euroqol Health State at two months, Caregiver Strain Index at two months and the Hospital Anxiety and Depression at two months. The time between stroke onset and admission to the stroke rehabilitation unit was also a retained factor, which is probably inherent to stroke severity but also part of stroke care/management in the different countries. Side of lesion was retained as prognostic factor in the psychosocial model, but post-hoc analyses could not identify significant differences. The specificity of the psychosocial dimension was lower (66%) compared to the other two dimensions (physical 85%, independence 84%). The prognosis for this model is thus less explanatory compared to the other two models. Lower specificity means that the model actually predicts patients with a better outcome in a lower category. SIP independent dimension Having hyperlipidaemia at intake was retained as an independent predictor in the model of the independent dimension of good participation at six months post stroke. We have no explanation for this finding. Again, the Barthel Index on admission, Euroqol Health State at two months and Caregiver Strain Index at two months were retained in the model and can be seen as the most important factors when looking at the improvement in participation outcome in general. Thus, it seems important to focus early stroke rehabilitation also on these variables. As CSI is retained in every model, there seems to be an interaction between strain of the caregiver and level of participation of the patient. This needs more in-depth investigation in future studies. The sensitivity in this model was lower (64%) compared to the other two models which reached values above 80%. This indicates that patients having an actual poor outcome were predicted to have a good outcome. The specificity value is 84%, comparable to the physical dimension (85%). Limitations Firstly, the aim of CERISE was not to measure and predict participation, but rather to compare recovery and rehabilitation after stroke across Europe. Salter et al. [22] concluded that

Predicting participation after stroke

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

DOI: 10.3109/09638288.2014.948137

‘‘the participation category seems to be the most problematic’’ with regard to measuring the same domains in different studies and the difficulty of having a mixture of measurements from other ICF categories. Particularly assessing participation and selecting the right outcome measure is complicated [22,33], because participation is always related to the other levels of the ICF. Salter et al. [22] stated that many existing measures include items of different ICF dimensions and also items which cannot be found in the ICF. This means that there are outcome measures which contain participation items as well as for example items of the Body Function/Structure dimension. Validated assessments of participation are still lacking and more research is needed in this field. Secondly, it has been suggested not to measure participation earlier than six months after stroke [14]. It is yet to be determined which time point should be targeted for predicting participation. People after stroke should probably be provided sufficient time to rebalance their level of participation after returning home from inpatient rehabilitation, or when being admitted to institutionalized care. Thirdly, the Caregiver Strain Index (CSI) remained in all models. To use the CSI you need to collect data from another person. This might make data collection more difficult and it might make using the presented prediction equation in practice more challenging. The present study is the largest European study which looked at predictors for participation at six months after stroke. The prediction models and equations can be used in clinical settings as guidance for stroke rehabilitation management, as they highlight variables, commonly assessed in rehabilitation, related to participation at six months after stroke. The models for the psychosocial and independent SIP dimensions are directly applicable for therapy management; however the model for the SIP physical dimension is limited to the centers used in the CERISE study.

6.

7.

8.

9.

10.

11.

12.

13. 14.

15.

16.

17.

18.

Acknowledgements This project was conducted by I. Baert (B), P. Berman (GB), H. Beyens (B), N. Brinkmann (D), L. Connell (GB), E. Dejaeger (B), W. De Weerdt (B), L. De Wit (B), H. Feys (B), W. Jenni (CH), J. Jurkat (D), H. Kamsteegt (B), C. Kaske (CH), M. Leys (B), N.B. Lincoln (GB), F. Louckx (B), K. Putman (B), B. Schuback (CH), W. Schupp (D) and B. Smith (GB).

19. 20. 21.

22.

Declaration of interest The authors report no conflicts of interest. This article was developed within the framework of the research ‘‘Collaborative Evaluation of Rehabilitation in Stroke across Europe (CERISE)’’, Quality of life-key action 6, 2001–2005, contract number QLK6-CT-2001-00170 funded by the European Commission and Bundesamt fu¨r Bildung und Wissenschaft (CH).

23.

References

25.

1. Inouye M. Predicting outcomes of patients in Japan after first acute stroke using a simple model. Am J Phys Med Rehabil 2001;80: 645–9. 2. Veerbeek JM, Kwakkel G, van Wegen EE, et al. Early prediction of outcome of activities of daily living after stroke: a systematic review. Stroke 2011;42:1482–8. 3. World Health Organization. International classification of functioning, disability and health: ICF. Geneva: World Health Organization; 2001. 4. Fiorelli M, Alperovitch A, Argentino C, et al. Prediction of longterm outcome in the early hours following acute ischemic stroke. Italian acute stroke study group. Arch Neurol 1995;52:250–5. 5. Johnston KC, Wagner DP, Wang XQ, et al. Validation of an acute ischemic stroke model: does diffusion-weighted imaging lesion

24.

26.

27.

28. 29.

947

volume offer a clinically significant improvement in prediction of outcome? Stroke 2007;38:1820–5. Reid JM, Gubitz GJ, Dai D, et al. Predicting functional outcome after stroke by modelling baseline clinical and CT variables. Age Ageing 2010;39:360–6. Sato S, Toyoda K, Uehara T, et al. Baseline NIH Stroke Scale Score predicting outcome in anterior and posterior circulation strokes. Neurology 2008;70:2371–7. Weimar C, Roth M, Willig V, et al. Development and validation of a prognostic model to predict recovery following intracerebral hemorrhage. J Neurol 2006;253:788–93. Weimar C, Ziegler A, Konig IR, Diener HC. Predicting functional outcome and survival after acute ischemic stroke. J Neurol 2002; 249:888–95. Sa´nchez-Blanco I, Ochoa-Sangrador C, Lo´pez-Munaı´n L, et al. Predictive model of functional independence in stroke patients admitted to a rehabilitation programme. Clin Rehabil 1999;13:464–75. Eyssen IC, Steultjens MP, Dekker J, Terwee CB. A systematic review of instruments assessing participation: challenges in defining participation. Arch Phys Med Rehabil 2011;92:983–97. Whiteneck G, Dijkers MP. Difficult to measure constructs: conceptual and methodological issues concerning participation and environmental factors. Arch Phys Med Rehabil 2009;90:S22–35. Mayo NE, Wood-Dauphinee S, Ahmed S, et al. Disablement following stroke. Disabil Rehabil 1999;21:258–68. Duncan PW, Jorgensen HS, Wade DT. Outcome measures in acute stroke trials: a systematic review and some recommendations to improve practice. Stroke 2000;31:1429–38. Mayo NE, Wood-Dauphinee S, Cote R, et al. Activity, participation, and quality of life 6 months poststroke. Arch Phys Med Rehabil 2002;83:1035–42. van der Zee CH, Visser-Meily JM, Lindeman E, et al. Participation in the chronic phase of stroke. Top Stroke Rehabil 2013;20:52–61. De Wit L, Putman K, Schuback B, et al. Motor and functional recovery after stroke: a comparison of 4 European rehabilitation centers. Stroke 2007;38:2101–7. The World Health Organization Monica Project (monitoring trends and determinants in cardiovascular disease): a major international collaboration. Who MONICA Project Principal Investigators. J Clin Epidemiol 1988;41:105–14. Lincoln N, Leadbitter D. Assessment of motor function in stroke patients. Physiotherapy 1979;65:48–51. Mahoney FI, Barthel DW. Functional evaluation: the Barthel index. Md State Med J 1965;14:61–5. Bergner M, Bobbitt RA, Carter WB, Gilson BS. The sickness impact profile: development and final revision of a health status measure. Med Care 1981;19:787–805. Salter K, Jutai JW, Teasell R, Foley NC, Bitensky J, Bayley M. Issues for selection of outcome measures in stroke rehabilitation: ICF participation. Disabil Rehabil 2005;27:507–28. Mock C. The sickness impact profile (SIP) in the assessment of the long-term functional outcome of an injured patient. In: Mock C, ed. The sickness impact profile (SIP) in the assessment of the long-term functional outcome of an injured patient. 3rd International Conference Measuring the Burden of Injury; 1994. Pollard WE, Bobbitt RA, Bergner M, et al. The sickness impact profile: reliability of a health status measure. Med Care 1976;14: 146–55. De Wit L, Putman K, Dejaeger E, et al. Use of time by stroke patients: a comparison of four european rehabilitation centers. Stroke 2005;36:1977–83. Putman K, de Wit L, Schupp W, et al. Use of time by physiotherapists and occupational therapists in a stroke rehabilitation unit: a comparison between four european rehabilitation centres. Disabil Rehabil 2006;28:1417–24. Kwakkel G, van Peppen R, Wagenaar RC, et al. Effects of augmented exercise therapy time after stroke: a meta-analysis. Stroke 2004:35:2529–39. Langhorne P, Wagenaar R, Partridge C. Physiotherapy after stroke: more is better? Physiother Res Int 1996;1:75–88. Putman K, De Wit L, Schupp W, et al. Variations in follow-up services after inpatient stroke rehabilitation: a multicentre study. J Rehabil Med 2009;41:646–53.

948

C. Stummer et al.

Disabil Rehabil, 2015; 37(11): 942–950

30. Clark MS, Smith DS. Psychological correlates of outcome following rehabilitation from stroke. Clin Rehabil 1999;13: 129–40. 31. Glass TA, Matchar DB, Belyea M, Feussner JR. Impact of social support on outcome in first stroke. Stroke 1993;24: 64–70.

32. Meijer R, Ihnenfeldt DS, de Groot IJ, et al. Prognostic factors for ambulation and activities of daily living in the subacute phase after stroke. A systematic review of the literature. Clin Rehabil 2003;17: 119–29. 33. Roberts L, Counsell C. Assessment of clinical outcomes in acute stroke trials. Stroke 1998;29:986–91.

Appendix

Table A. Univariate logistic regression results for SIP physical dimension (good outcome 520%).

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

Variable Center: Pell, Nott, Zurz, Herz Sex: M/F Side of lesion: L/R/B/U Type of stroke: isch./haem./U Neurosurgical intervention: y/n Myocardial infarction: y/n Diabetes Mellitus: y/n History of high blood pressure: y/n Ever smoked: y/n Atrial fibrillation: y/n Chronic heart disease: y/n Hyperlipidaemia: y/n Urinary incontinence: y/n Secondary complications: y/n Pre-Stroke Rankin (range: 0–5) Swallowing problem: y/n TSOA (days) Age (yr) Pre stroke BI (range: 55–100) NIHSS (range: 0–23) RMA-GF (range: 0–13) RMA-LT (range: 0–10) RMA-A (range: 0–15) BI (range: 0–100) Monthly income (1, 2, 3, U) Education (1, 2, U) RMA-GF 2 RMA-LT 2 RMA-A 2 BI 2 EADL 2 EQHS 2 EQVAS 2 HADS 2 CSI 2 RMA-GF S RMA-LT S RMA-A S BI S

B

Sig.

Exp (B)

95%CI for Exp (B)

0.55 0.69 0.13 0.43 0.14 0.07 0.11 0.07 0.31 1.01 0.19 0.41 1.11 0.60 0.66 1.20 0.03 0.05 0.17 0.20 0.34 0.41 0.16 0.05 0.38 0.91 0.60 0.67 0.22 7.01 0.23 0.08 0.04 0.22 0.19 0.15 0.15 0.20 0.01

0.00 0.00 0.41 0.10 0.77 0.81 0.66 0.74 0.12 0.00 0.39 0.04 0.00 0.02 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.34

0.58 1.99 1.14 0.65 1.15 1.07 1.11 0.93 0.74 2.75 1.21 0.67 .33 1.82 1.94 3.33 1.03 1.05 0.84 1.22 0.71 0.67 0.85 0.95 0.68 0.41 0.55 0.51 0.81 0.92 0.79 0.93 0.96 1.25 1.21 0.86 0.86 0.82 1.01

0.48–0.69 1.35–2.93 0.84–1.54 0.39–1.08 0.45–2.98 0.61–1.90 0.69–1.78 0.62–1.40 0.50–1.09 1.59–4.77 0.78–1.89 0.46–0.98 0.19–0.58 1.09–3.06 1.41–2.66 1.86–5.97 1.01–1.05 1.03–1.07 0.76–0.94 1.15–1.28 0.67–0.76 0.61–0.72 0.81–0.89 0.94–0.96 0.54–0.86 0.26–0.63 0.49–0.62 0.45–0.59 0.77–0.84 0.91–0.94 0.76–0.83 0.91–0.94 0.95–0.98 1.17–1.33 1.13–1.30 0.80–0.93 0.78–0.94 0.76–0.88 1.00–1.02

Values in bold are significant at the level of p50.1. SIP indicates Sicknes Impact Profile; Pell, Pellenberg; Nott, Nottingham; Zurz, Zurzach; Herz, Herzogenaurach; M/F, male/female; L/R/B/U, left/right/both/unknown; isch./haem/U, ischemic/haemorrhage/unknown; y/n, yes/no; TSOA, time between stroke onset and admission to the stroke rehabilitation unit in days; yr, years; BI, Barthel Index; NIHSS, National Institutes of Health Stroke Scale; RMA-GF, Rivermead Motor assessment-Gross motor function; RMA-LT, Rivermead Motor assessment-leg and trunk; RMA-A, Rivermead Motor assessment-arm; Monthly income: 1 ¼ low income (560% of median national equivalent income) 2 ¼ moderate (60–120%) 3 ¼ high (4120%) U ¼ unknown; Education: 1 ¼ low education (below or equal to secondary level), 2 ¼ high education (upper secondary level or higher) U ¼ unknown; CI, confidence interval; RMA-GF 2, Rivermead Motor assessment-Gross motor function at two months (range: 0–13); RMA-LT 2, Rivermead Motor assessment-leg and trunk at two months (range: 0–10); RMA-A 2, Rivermead Motor assessment-arm at two months (range: 0–15); EADL 2, Extended Activities of Daily Living at two months (range: 0–22); CI, confidence interval; EQHS2, Euroqol Health State at 2 months; EQVAS 2, Euroqol-Visual analogue scale at 2 months; HADS, Hospital Anxiety and Depression Scale (range: 0–42); CSI, Caregiver Strain Index (range: 0–13); S, two months variable minus intake variable; N range: 444–532.

Predicting participation after stroke

DOI: 10.3109/09638288.2014.948137

Table B. Univariate logistic regression results for SIP psychosocial dimension (good outcome 520%).

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

Variable Center, Pell, Nott, Zurz, Herz Sex M/F Side of lesion: L/R/B/U Type of stroke: isch./haem./U Neurosurgical intervention: y/n Myocardial infarction: y/n Diabetes Mellitus: y/n History of high blood pressure: y/n Ever smoked: y/n Atrial fibrillation: y/n Chronic heart disease: y/n Hyperlipidaemia: y/n Urinary incontinence: y/n Secondary complications: y/n Pre-Stroke Rankin (range: 0–5) Swallowing problem: y/n TSOA (days) Age (yr) Pre stroke BI (range: 55–100) NIHSS (range: 0–23) RMA-GF (range: 0–13) RMA-LT (range: 0–10) RMA-A (range: 0–15) BI (range: 0–100) Monthly income (1, 2, 3, U) Education (1, 2, U) RMA-GF 2 RMA-LT 2 RMA-A 2 BI 2 EADL 2 EQHS 2 EQVAS 2 HADS 2 CSI 2 RMA-GF S RMA-LT S RMA-A S BI S

B

Sig.

Exp (B)

95%CI for Exp (B)

0.28 0.39 0.28 0.09 0.39 0.14 0.04 0.00 0.27 0.30 0.16 0.38 0.29 0.38 0.53 1.03 0.04 0.02 0.12 0.12 0.15 0.17 0.08 0.03 0.26 0.60 0.19 0.22 0.11 0.03 0.16 0.05 0.03 0.21 0.15 0.15 0.17 0.15 0.00

0.00 0.04 0.07 0.73 0.41 0.63 0.86 0.99 0.17 0.22 0.47 0.05 0.23 0.12 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.53

0.76 1.48 0.76 0.92 1.47 1.15 1.04 1.00 0.76 1.35 0.85 0.68 0.75 1.47 1.70 2.79 1.04 1.02 0.89 1.13 0.86 0.84 0.92 0.98 0.77 0.55 0.83 0.80 0.90 0.97 0.85 0.95 0.97 1.23 1.16 0.86 0.84 0.86 1.00

0.64–0.90 1.01–2.16 0.56–1.03 0.55–1.52 0.59–3.69 0.65–2.01 0.65–1.67 0.67–1.48 0.52–1.12 0.84–2.16 0.55–1.32 0.47–1.00 0.46–1.20 0.91–2.37 1.31–2.20 1.70–4.58 1.02–1.06 1.01–1.04 0.83–.95 1.09–1.18 0.82–0.91 0.79– 0.90 0.89–0.96 0.97–0.98 0.62–0.96 0.36–0.85 0.79–0.87 0.75– 0.85 0.87–0.93 0.96–0.98 0.82–0.88 0.94–0.96 0.96–0.98 1.16–1.30 1.09–1.23 0.80–0.93 0.77–0.93 0.80–0.92 0.99–1.01

Values in bold are significant at the level of p50.1. SIP indicates Sicknes Impact Profile; Pell, Pellenberg; Nott, Nottingham; Zurz, Zurzach; Herz, Herzogenaurach; M/F, male/female; L/R/B/U, left/right/both/unknown; isch./haem/U, ischemic/haemorrhage/unknown; y/n, yes/no; TSOA, time between stroke onset and admission to the stroke rehabilitation unit in days; yr, years; BI, Barthel Index; NIHSS, National Institutes of Health Stroke Scale; RMA-GF, Rivermead Motor assessment-Gross motor function; RMA-LT, Rivermead Motor assessment-leg and trunk; RMA-A, Rivermead Motor assessment-arm; Monthly income: 1 ¼ low income (560% of median national equivalent income) 2 ¼ moderate (60–120%) 3 ¼ high (4120%) U ¼ unknown; Education: 1 ¼ low education (below or equal to secondary level), 2 ¼ high education (upper secondary level or higher) U ¼ unknown; CI, confidence interval; RMA-GF 2, Rivermead Motor assessment-Gross motor function at two months (range: 0–13); RMA-LT 2, Rivermead Motor assessment-leg and trunk at two months (range: 0–10); RMA-A 2,Rivermead Motor assessment-arm at two months (range: 0–15); EADL 2, Extended Activities of Daily Living at two months (range: 0–22); CI, confidence interval; EQHS2, Euroqol Health State at 2 months; EQVAS 2, Euroqol-Visual analogue scale at 2 months; HADS, Hospital Anxiety and Depression Scale (range: 0–42); CSI, Caregiver Strain Index (range: 0–13); S, two months variable minus intake variable; N range: 444–532.

949

950

C. Stummer et al.

Disabil Rehabil, 2015; 37(11): 942–950

Table C. Univariate logistic regression results for SIP independent dimension (good outcome 520%).

Disabil Rehabil Downloaded from informahealthcare.com by Nanyang Technological University on 04/24/15 For personal use only.

Variable Center, Pell, Nott, Zurz, Herz Sex M/F Side of lesion: L/R/B/U Type of stroke: isch./haem./U Neurosurgical intervention: y/n Myocardial infarction: y/n Diabetes Mellitus: y/n History of high blood pressure: y/n Ever smoked: y/n Atrial fibrillation: y/n Chronic heart disease: y/n Hyperlipidaemia: y/n Urinary incontinence: y/n Secondary complications: y/n Pre-Stroke Rankin (range: 0–5) Swallowing problem: y/n TSOA (days) Age (yr) Pre stroke BI (range: 55–100) NIHSS (range: 0–23) RMA-GF (range: 0–13) RMA-LT (range: 0–10) RMA-A (range: 0–15) BI (range: 0–100) Monthly income (1, 2, 3, U) Education (1, 2, U) RMA-GF 2 RMA-LT 2 RMA-A 2 BI 2 EADL 2 EQHS 2 EQVAS 2 HADS 2 CSI 2 RMA-GF S RMA-LT S RMA-A S BI S

B

Sig.

Exp (B)

95%CI for Exp(B)

0.34 0.58 0.07 0.13 1.22 0.04 0.19 0.44 0.19 0.60 0.03 0.55 0.70 0.83 0.23 1.13 0.02 0.02 0.07 0.18 0.27 0.33 0.16 0.04 0.27 0.68 0.37 0.43 0.20 0.05 0.18 0.06 0.03 0.17 0.16 0.09 0.08 0.16 0.01

0.00 0.00 0.68 0.62 0.05 0.90 0.43 0.04 0.34 0.03 0.91 0.01 0.01 0.00 0.09 0.00 0.06 0.11 0.06 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.08 0.00 0.13

0.71 1.79 1.07 0.88 3.40 1.04 1.22 0.65 0.83 1.83 1.03 0.58 0.50 2.30 1.25 3.11 1.02 1.02 0.94 1.20 0.76 0.72 0.85 0.96 0.77 0.51 0.69 0.65 0.82 0.95 0.84 0.94 0.97 1.19 1.17 0.91 0.92 0.85 1.01

0.60–.85 1.21–2.65 0.79–1.45 0.53–1.47 0.98–11.78 0.58–1.86 0.75–1.98 0.43–.98 0.56–1.22 1.08–3.10 0.66–1.60 0.40–.85 0.30–0.85 1.32–3.98 0.97–1.63 1.71–5.64 1.00–1.04 1.00–1.03 0.87–1.00 1.14–1.26 0.72–0.81 0.66–0.77 0.82–0.89 0.95–0.97 0.61–0.96 0.33–0.77 0.64–0.75 0.59–0.72 0.78–0.85 0.94–0.97 0.81–0.87 0.93–0.95 0.96–0.98 1.12–1.26 1.10–1.25 0.85–0.98 0.84–10.01 0.80–0.91 1.00–1.02

Values in bold are significant at the level of p50.1. SIP indicates Sicknes Impact Profile; Pell, Pellenberg; Nott, Nottingham; Zurz, Zurzach; Herz, Herzogenaurach; M/F, male/female; L/R/B/U, left/right/both/unknown; isch./haem/U, ischemic/haemorrhage/unknown; y/n, yes/no; TSOA, time between stroke onset and admission to the stroke rehabilitation unit in days; yr, years; BI, Barthel Index; NIHSS, National Institutes of Health Stroke Scale; RMA-GF, Rivermead Motor assessment-Gross motor function; RMA-LT, Rivermead Motor assessment-leg and trunk; RMA-A, Rivermead Motor assessment-arm; Monthly income: 1 ¼ low income (560% of median national equivalent income) 2 ¼ moderate (60–120%) 3 ¼ high (4120%) U ¼ unknown; Education: 1 ¼ low education (below or equal to secondary level), 2 ¼ high education (upper secondary level or higher) U ¼ unknown; CI, confidence interval; RMA-GF 2, Rivermead Motor assessment-Gross motor function at two months (range: 0–13); RMA-LT 2, Rivermead Motor assessment-leg and trunk at two months (range: 0–10); RMA-A 2,Rivermead Motor assessment-arm at two months (range: 0–15); EADL 2, Extended Activities of Daily Living at two months (range: 0–22); CI, confidence interval; EQHS2, Euroqol Health State at 2 months; EQVAS 2, Euroqol-Visual analogue scale at 2 months; HADS, Hospital Anxiety and Depression Scale (range: 0–42); CSI, Caregiver Strain Index (range: 0–13); S, two months variable – intake variable; N range: 444–532.