Motor and Functional Recovery After Stroke
A Comparison of 4 European Rehabilitation Centers
Background and Purpose— Outcome after first stroke varies significantly across Europe. This study was designed to compare motor and functional recovery after stroke between four European rehabilitation centers.
Methods— Consecutive stroke patients (532 patients) were recruited. They were assessed on admission and at 2, 4, and 6 months after stroke with the Barthel Index, Rivermead Motor Assessment of Gross Function, Rivermead Motor Assessment of Leg/Trunk, Rivermead Motor Assessment of Arm, and Nottingham Extended Activities of Daily Living (except on admission). Data were analyzed using random effects ordinal logistic models adjusting for case-mix and multiple testing.
Results— Patients in the UK center were more likely to stay in lower Rivermead Motor Assessment of Gross Function classes compared with patients in the German center (ΔOR, 2.4; 95% CI, 1.3 to 4.3). In the Swiss center, patients were less likely to stay in lower Nottingham Extended Activities of Daily Living classes compared with patients in the UK center (ΔOR, 0.7; 95% CI, 0.5 to 0.9). The latter were less likely to stay in lower Barthel Index classes compared with the patients in the German center (ΔOR, 0.6; 95%CI, 0.4 to 0.8). Recovery patterns of Rivermead Motor Assessment of Leg/Trunk and Rivermead Motor Assessment of Arm were not significantly different between centers.
Conclusions— Gross motor and functional recovery were better in the German and Swiss centers compared with the UK center, respectively. Personal self-care recovery was better in the UK compared with the German center. Previous studies in the same centers indicated that German and Swiss patients received more therapy per day. This was not the result of more staff but of a more efficient use of human resources. This study indicates potential for improving rehabilitation outcomes in the UK and Belgian centers.
Stroke is a major health burden throughout Europe, consuming significant resources.1 Multidisciplinary rehabilitation in inpatient stroke rehabilitation units results in lower mortality, less disability, and less need for institutionalization compared with care on general wards.2 Previous studies have shown that outcome after inpatient stroke rehabilitation differs widely across Europe.3,4 Overall mortality and dependency rates were higher in the United Kingdom than in other European countries.3,4 Weir et al4 assumed that differences were too large to be caused by variation in stroke management and attributed them to unmeasured case-mix variables.
Despite the evidence that organized care in stroke rehabilitation units is associated with improved outcome,2 we have limited knowledge of the components of such care that are responsible for this benefit.1 Comparing inpatient stroke care and recovery patterns across different European countries might improve our understanding of stroke rehabilitation and may help develop optimal models for delivery of stroke care. However, most comparative studies measure outcome only in terms of dependency or death at a fixed time after stroke.3,4 Such studies do not provide insight into the time course of recovery.
Adjustment for case-mix is required to enable valid comparison of recovery patterns between different groups.5 Prognostic models for recovery after stroke have been developed, resulting in a long list of case-mix variables.6–9 Age, aphasia and initial impairment have been associated with motor recovery.9 Tilling et al8 found that the functional recovery of patients admitted for rehabilitation was predicted by urinary incontinence, sex, prestroke disability, dysarthria, age, dysphasia, and limb deficit.
The CERISE project was a multicenter longitudinal cohort study comparing inpatient stroke care and recovery patterns between four European rehabilitation centers. The aim was to assess variation in motor and functional recovery patterns for 6 months after stroke.
Patients and Methods
Subjects and Settings
The study was conducted in 4 European rehabilitation centers: University Hospital, Leuven, Belgium; City Hospital and Queen’s Medical Centre, Nottingham, United Kingdom; RehaClinic, Zurzach, Switzerland; and Fachklinik, Herzogenaurach, Germany. Each center had a stroke rehabilitation unit with the provision of inpatient multidisciplinary care. Patients were transferred to these units from an acute setting. Between March 2002 and September 2004, all consecutive patients fitting the following inclusion criteria were recruited: (1) first-ever stroke as defined by WHO10; (2) score10 on Rivermead Motor Assessment11 of Gross Function (RMA-GF) 11, and/or score on Rivermead Motor Assessment of Leg/Trunk (RMA-LT) function 8 and/or score on Rivermead Motor Assessment of Arm (RMA-A) function 12 on admission to the center; and (3) age 40 to 85 years. The upper age limit was to avoid inclusion of patients with a high number of comorbidities. 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 >6 weeks after stroke (to exclude chronic stroke patients); (4) no informed consent; and (5) prestroke Barthel Index (BI)12 <50 (to be able to distinguish between pre-existing disabilities and disabilities resulting from the stroke). The study was approved by the ethics committee for each center.
Motor and functional recovery were assessed on admission to the center and at 2, 4, and 6 months after stroke with the RMA-GF, RMA-LT, RMA-A, and BI, respectively. The Nottingham Extended Activities of Daily Living (NEADL)13 was assessed 2, 4, and 6 months after stroke to evaluate independence in instrumental activities of daily living. The end point was 6 months after stroke as most motor and functional recovery takes place before that time. To enable adjustment for case-mix, several variables were documented based on previous studies of prognostic factors for motor9 and functional recovery.6–8 Age,6–9 gender,8 time between stroke onset and admission assessment, prestroke disability8 (assessed by BI), type of stroke, side of impairment, urinary incontinence,6–8 and swallowing problems6,7 were recorded on admission. The occurrence of dysarthria8 and dysphasia8,9 were documented using items of the National Institute of Health Stroke scale.14 At discharge, length of stay was recorded.
A researcher in each center collected all data. The researchers were trained in the assessments at the onset of the study. A manual was provided to ensure standardization. The project manager (L.D.W.) visited each center four times to recalibrate the researchers’ work.
Baseline and demographic patients’ characteristics were compared between centers using χ2, ANOVA, or Kruskal-Wallis tests, as appropriate. Differences between centers were determined with post-hoc tests.
Comparison of the recovery patterns across centers over time required adjustment for case-mix and for missing data. Furthermore, the distribution of the outcome measurements was skewed. The skewness differed between centers and between evaluation points which prevented transformation to a normal distribution. The time between stroke onset and admission assessment was significantly shorter in the UK center compared with the other centers (12 days versus ≈3 weeks). The combination of the aforementioned statistical problems made it impossible to use conventional statistics, such as ANOVA or linear mixed models. Therefore, data were analyzed using random effects ordinal logistic models with random intercept and slope.15,16 In this model, the logarithm of the odds (proportion of cases in low-response versus high-response classes) was expressed as a linear function of covariates and random effects. This model allowed comparison of the odds ratios across centers. For the analysis, the 5 response measurements were divided in classes based on their distribution. The RMA-GF was divided into 5 classes (0–2, 3–5, 6–7, 8–9, 10–13); RMA-LT into 4 classes (0–3, 4–6, 7–9,10); RMA-A into 5 classes (0–1, 2–5, 6–9, 10–12, 13–15); BI into 5 classes (0–20, 25–40, 45–60, 65–80, 85–100); and NEADL into 6 classes (0–2, 3–5, 6–8, 9–11, 12–16, 17–22). Age, gender, urinary incontinence, swallowing problems, dysphasia, and dysarthria were evaluated as confounders for each outcome measure and compared between centers and over time. To adjust for differences in time between stroke onset and admission, patient score at 3 weeks was estimated in the model as the baseline for all centers.
For each outcome measure, a full model was built in which the logarithm of the odds was regressed on time, center, time*center, all potential confounders, and their interaction with time and center, respectively. All confounders and interaction effects that were not significant at P<0.05 were subsequently omitted from the model. Age and gender were always forced into the final model because both variables were significantly different between centers resulting in imbalanced groups and, from a population perspective, both variables influence almost all body functions and therefore should be included even though they appear not to be significant confounders in the study sample.
A major confounder for recovery was the severity of the stroke. Severity was reflected in patients’ initial motor and functional deficits. The latter could not be included as confounders in the final model because the motor and functional scores were also the outcomes and, in the UK center, the baseline assessment took place earlier than in the other centers. However, by including the baseline value as the first response and by choosing the interaction term “time*center” as a measure for the effect of center on the time course of recovery, the initial difference in stroke severity between centers was accounted for.
Significant effects of “time*center” were further explored via pair-wise center comparisons over time to evaluate which centers had a significantly different evolution of the outcome measure over time. This resulted in the rate of change of OR per unit of time, where unit of time was set at 1 month. Holm’s method was used to correct for multiple comparisons.17 Statistical analyses were performed with SAS (version 8.2). For the random effects ordinal logistic models, OpenBUGS (version 2.1.0) and R (version 2.2.0) were used.
Between March 2002 and September 2004, 1297 stroke patients were admitted to the 4 centers. Of these, 765 patients were excluded (no/minor motor impairment, n=275; admitted to the center >6 weeks after stroke, n=95; assessment >5 days after admission, n=31; other neurological impairment, n=23; stroke-like symptoms caused by trauma, n=15; refused consent, n=275; no inclusion for practical reasons, n=54). The remaining 532 were included in the study. Figure 1 shows the numbers lost to follow-up at the different time points. At 6 months after stroke, 69 patients were lost to follow-up; 18 died, 46 refused to participate, and 5 could not be assessed (missed assessment, poor medical condition).
Demographic and prognostic data were compared between centers (Table 1). Patients in the UK and Swiss centers were significantly older than those in the Belgian and German centers. In the German center, there were significantly more male patients compared with the other centers. Time between stroke onset and admission assessment and length of stay were significantly shorter in the UK center compared with the other 3 centers. Length of stay was also significantly shorter in the German center compared with the Belgian and Swiss centers. No significant differences were found between centers for side of impairment, type of stroke, or prestroke BI. Comparison of prognostic data between centers revealed that in the Belgian and UK centers, there were significantly more patients with urinary incontinence. Patients in the UK center also had significantly more swallowing problems, whereas dysarthria occurred significantly more in the Belgian center and dysphasia more in the Swiss center. Patients in the Swiss and German centers had significantly less severe strokes compared with the other 2 centers, reflected in higher initial BI, RMA-GF, RMA-LT, and RMA-A scores.
Comparison of Motor and Functional Recovery Patterns Between Centers
The different patient profiles on admission required case-mix adjustment. The final models containing the significant confounders for each outcome measurement are presented in Table 2. When fitting these models, time*center was a significant interaction term for RMA-GF, BI, and NEADL. This indicates that the recovery patterns of RMA-GF, BI, and NEADL were significantly different between centers. For RMA-LT and RMA-A, time*center was not significant.
Comparison between centers (Table 3, Figure 2) by means of the rate of change in OR over time showed that patients in the UK center were significantly more likely to stay in lower RMA-GF classes compared with patients in the German center. Patients in the UK center were significantly less likely to stay in lower BI classes compared with patients in the German center. Patients in the Swiss center were significantly less likely to stay in lower NEADL classes compared with patients in the UK center.
Comparison of motor and functional recovery between four European rehabilitation centers for six months after stroke revealed significantly different recovery patterns for RMA-GF, BI and NEADL. No significant differences were found in the recovery patterns of RMA-LT and RMA-A. Within the CERISE project other studies were conducted to compare stroke care between these centers. The main results are summarized below to help understand the present findings.
First, a time sampling study18 showed an average daily therapy time of 1 hour in the UK, 2 hours in the Belgian, 2 hours 20 minutes in the German, and 2 hours 46 minutes in the Swiss center. In all centers, physiotherapy comprised nearly 40% of therapeutic time. Occupational therapy comprised 20% to 30%, except in the UK center (11.6%). In the latter, 35% of therapy time consisted of nursing care. Lying and sitting occupied almost 5 hours in the UK, 3.5 hours in the Belgian, and <3 hours in the Swiss and German centers. Patients in the German center spent more time in leisure activities compared with those in the other centers. After correction for case-mix, overall therapy time in the UK center was significantly less than the other centers. Also, occupational therapy time was significantly less in the UK compared with the Swiss center. Differences in therapy time were not attributable to differences in patients/staff ratio.18
In a second study,19 the content of physiotherapy and occupational therapy were compared between the centers. The content was similar in all 4 centers. In a third study,20 we compared the activities of physical and occupational therapists. The proportion of time spent on direct patient care was the highest for the German physical (66.1%) and occupational therapists (63.3%) and lowest in the UK center, 45.9% and 32.9%, respectively. Therapists in the UK center spent more than half of their time on nontherapeutic activities (administrative tasks, ward rounds, etc). This resulted in less time for direct patient care. In the German and Swiss centers, the rehabilitation programs were strictly timed, while in the Belgian and UK centers they were organized on an ad hoc basis. The results of previous studies18–20 showed significant differences in therapy time and task divisions between centers. No differences were found in the content of physiotherapy and occupational therapy. The more formal management in the German center may have resulted in the most efficient use of human resources, which may have resulted in more therapy time for the patients.
In the present study, we found that patients in the UK center were significantly more likely to stay in lower RMA-GF and NEADL classes compared with the patients in the German and Swiss center, respectively. It seems that the higher input of therapy in the German and Swiss centers18 resulted in a better gross motor and functional recovery compared with the UK center. The fact that better NEADL recovery was found in the Swiss center may be attributable to significantly higher occupational therapy input in the Swiss compared with the UK center. A significantly better gross motor recovery (RMA-GF) was found only in the German center. Patients in the German center spent least time in passive and most time in leisure activities: it may be that a more active atmosphere contributed to a better motor recovery.
In contrast to the RMA-GF and NEADL results, patients in the UK center were less likely to stay in lower BI classes compared with patients in the German center. BI reflects patients’ dependency and need for supervision in personal self-care.12 Patients in the UK center received more nursing compared with patients in the other centers.18 Nurses are an integral part of the stroke rehabilitation team. They reinforce the rehabilitation strategies and help to achieve the greatest personal independence possible.21 Hence, the high input of nursing care in the UK center may have contributed to the better BI recovery. Second, in the UK center, the median BI score on admission was 45% of the total BI range. It has been reported that patients with middle band scores (30% to 60% of maximum score) can expect the most functional gain.22 However, patients in the Belgian center also had middle band BI scores on admission (40% of BI range), but did not show significantly better BI recovery. Another reason may be that length of stay was the shortest in the UK (median=44.5 days) and longest in the Belgian center (median=66 days). The optimal BI recovery in the UK center may therefore reflect an emphasis on personal self-care to enable early discharge. This finding also suggests that discharge arrangements may affect patients’ recovery.
Poor gross motor recovery (RMA-GF) was in contrast to the good personal self-care recovery (BI) in the UK center. However, good personal self-care recovery (BI) was in contrast to poor NEADL recovery. The initial training in Activities of Daily Living (ADL) emphasized compensatory strategies,23 eg, self-care was performed with 1-hand techniques by using the unaffected arm. Hence, patients in the UK center may have improved in self-care (BI) through compensation with the unaffected side, while the actual motor deficit remained unchanged. Compared with the BI, the NEADL evaluates dependency at a higher level, requiring a certain level of motor recovery. These findings suggest that the NEADL is a better outcome measure than the BI for measuring the effects of rehabilitation.
The recovery patterns of patients in the Belgian center did not differ significantly from patients in any other center. Also, there were no significant differences between centers in the recovery patterns of RMA-LT or RMA-A. However, without adjustment for multiple testing, results showed that recovery of RMA-LT was worse in the UK center compared with the German center (P=0.03) and recovery of RMA-A was worse in the UK center compared with the Swiss center (P=0.04). These findings are in line with the results above.
Critical reflections on this study are appropriate. First, the centers were selected because of their established reputation for stroke rehabilitation. Generalization of the results to the country is not appropriate as only one center per country was included. Second, 275 stroke patients refuse to participate. Their admission profile could not be recorded for ethical reasons. Third, in all centers, the majority of patients were discharged between 2 and 4 months after stroke. At 4 and 6 months after stroke, recovery may be affected by the care after discharge. However, at discharge, half of the patients reached a score of at least 70% of the maximum RMA-GF score and at least 85% of the maximum BI score. Consequently, most recovery took place during inpatient rehabilitation, indicating that this period was crucial in the recovery phase. A major problem was the timing of baseline assessment and the difference between centers with respect to the prognostic factors. Such differences do not enable a straightforward comparison of recovery patterns. We therefore applied the random effects ordinal logistic models that estimated scores at identical time points in all centers and adjusted for case-mix. By calculating the rate of change of OR in time, the motor and functional deficits at baseline were taken into account. Other (unmeasured) factors may have also affected recovery (eg, patients’ motivation, therapists’ experience, etc). Because of statistical restrictions, it was impossible to consider them all. We based case-mix variables on high-quality studies8,9 and reviews.6,7 Although the number of variables included in the models was high, we believe that data are not overfitted, because a minimum 10 patients/included variable was achieved. Furthermore, when applying complex statistical models, power calculation (a priori/posthoc) is problematic. To obtain an idea of the true difference, the 95% CI of the rate of change of OR is a valid alternative. Despite the application of sophisticated statistical models, the authors are aware that a statistical model cannot correct for the fact that the more severely affected patients in the UK and Belgian centers (Table 1) might have less rehabilitation potential. However, patients in the Swiss and UK centers were significantly older, also indicating less rehabilitation potential. Dysphasia occurred more in the Swiss and dysarthria more in the Belgian center. Although individual prognostic factors differed significantly, the alternating direction of the differences resulted in similar patient profiles with respect to rehabilitation potential. Within the limits of statistical possibilities, we believe that we have made a serious attempt to correct for case-mix, but we cannot guarantee that no residual confounding occurred. Therefore, the results must be interpreted with caution.
Despite these methodological issues, our findings are in line with the results of a meta-analysis24 and a review,25 indicating that more intensive rehabilitation results in better recovery. Meta-analyses and reviews are based on randomized controlled trials. These experimental designs generate an artificial situation that may jeopardize generalization of the results.26 The advantage of this observational study was that we compared the real situation in existing settings. Further studies are needed to confirm the findings of the CERISE project. Ideally, centers should be compared which have equivalent patient groups on admission, but fundamental differences in patient management. Such studies will provide insight into the aspects of stroke rehabilitation that are crucial for patients’ outcome. This will provide better guidelines to improve stroke rehabilitation, which in turn will lead to less long-term disability.
The results showed that gross motor and functional recovery were better in the Swiss and German centers than in the UK center, with the exception of personal self-care recovery in the UK. In the German and Swiss centers, patients received more therapy per day. This was not the result of more staff but of a more efficient organization of rehabilitation services. This study indicated a potential for further improvement of the outcomes of rehabilitation in the UK and Belgian centers.
Sources of Funding
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 Sekretariat für Bildung und Forschung SBF (C.H.). It does not necessarily reflect its views and in no way anticipates the Commission’s future policy in this area.
- Received January 19, 2007.
- Accepted February 26, 2007.
Markus HS. Variations in care and outcome in the first year after stroke: a Western and Central European perspective. J Neurol Neurosurg Psychiatry. 2004; 75: 1660–1661.
Stroke Unit Trialists’ Collaboration. Organised inpatient (stroke unit) care for stroke. Cochrane Database Syst Rev. 2002.
Wolfe CDA, Tilling K, Rudd A, Giroud M, Inzitari D. Variations in care and outcome in the first year after stroke: a Western and Central European perspective. J Neurol Neurosurg Psychiatry. 2004; 75: 1702–1706.
Weir NU, Sandercock PAG, Lewis SC, Signorini DF, Warlow CP; on behalf of the IST collaborative group. Variations between countries in outcome after stroke in the International Stroke Trial (IST). Stroke. 2001; 322: 1370–1377.
Davenport RJ, Dennis MS, Warlow CP. Effect of correcting outcome data for case mix: an example from stroke medicine. BMJ. 1996; 312: 1503–1505.
Kwakkel G, Wagenaar RC, Kollen BJ, Lankhorst GJ. Predicting disability in stroke: a critical review of the literature. Age Ageing. 1996; 25: 479–489.
Meijer R, Ihnenfeldt DS, de Groot IJM, van Limbeek J, Vermeulen M, de Haan RJ. 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: 119–129.
Tilling K, Sterne JA, Rudd AG, Glass TA, Wityk RJ, Wolfe CD. A new method for predicting recovery after stroke. Stroke. 2001; 32: 2867–2873.
Nouri FM, Lincoln NB. An extended activity of daily living scale for stroke patients. Clin Rehabil. 1978; 1: 301–305.
Brott T, Adams HP, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Hertzberg V. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989; 20: 864–870.
Molenberghs G, Verbeke G. Models for discrete longitudinal data. New York: Springer; 2005.
Agresti A. Categorical Data Analysis, 2nd ed. Hoboken: John Wiley and Sons; 2002.
Holm S. A simple sequentially rejective multiple test procedure. Scand J Stat. 1979; 6: 65–70.
De Wit L, Putman K, Dejaeger E, Baert I, Berman P, Bogaerts K, Brinkmann N, Connell L, Feys H, Jenni W, Kaske C, Lesaffre E, Leys M, Lincoln N, Louckx F, Schuback B, Schupp W, Smith B, De Weerdt W. Use of time by stroke patients. A comparison of four European rehabilitation centers. Stroke. 2005; 36: 1977–1983.
De Wit L, Putman K, Lincoln N, Baert I, Berman P, Beyens H, Bogaerts K, Brinkmann N, Connell L, Dejaeger E, De Weerdt W, Jenni W, Lesaffre E, Leys M, Louckx F, Schuback B, Schupp W, Smith B, Feys H. Stroke rehabilitation in Europe. What do physiotherapists and occupational therapists actually do? Stroke. 2006; 37: 1483–1489.
Putman K, De Wit L, Schupp W, Baert I, Berman P, Connell L, Dejaeger E, De Meyer A, De Weerdt W, Feys H, Jenni W, Lincoln N, Louckx F, Martens A, Schuback B, Smith B, Leys M. Use of time by physiotherapists and occupational therapists in a stroke rehabilitation unit: a comparison between four European rehabilitation centres. Disabil Rehabil. 2006; 28: 1417–1424.
Alexander MP. Stroke rehabilitation outcome. A potential use of predictive variables to establish levels of care. Stroke. 1994; 25: 128–134.
Kwakkel G, van Peppen R, Wagenaar RC, Wood Dauphinee S, Richards C, Ashburn A, Miller K, Lincoln N, Partridge C, Wellwood I, Langhorne P. Effects of augmented exercise therapy time after stroke: a meta-analysis. Stroke. 2004; 35: 2529–2539.