SEVERAL STUDIES on stroke recovery have been conducted

Transcription

1 695 Early Versus Delayed Inpatient Stroke Rehabilitation: A Matched Comparison Conducted in Italy Stefano Paolucci, MD, Gabriella Antonucci, PhD, Maria Grazia Grasso, MD, PhD, Daniela Morelli, MD, Elio Troisi, MD, Paola Coiro, MD, Maura Bragoni, MD, PhD ABSTRACT. Paolucci S, Antonucci G, Grasso MG, Morelli D, Troisi E, Coiro P, Bragoni M. Early versus delayed inpatient stroke rehabilitation: a matched comparison conducted in Italy. Arch Phys Med Rehabil 2000;81: Objective: To assess the specific influence of onset admission interval () on rehabilitation results. Design: A case-control study in consecutive stroke inpatients, enrolled in homogeneous subgroups, matched for age (within 1 year) and Barthel Index (BI) score at admission, and different for to the rehabilitation ward. The short group began rehabilitation treatment within the first 20 days from stroke, medium group between days 21 and 40, and long between days 41 and 60. Setting: Rehabilitation hospital. Patients: One hundred forty-five patients with sequelae of first stroke. Main Outcome Measures: Efficiency (average increase in BI per day), effectiveness (proportion of potential improvement achieved during rehabilitation) of treatment, and percentage of low- and high-response patients, calculated on BI, were evaluated. Odds ratios (ORs) of dropouts and of poor and excellent therapeutic response were also quantified. Results: The short subgroup had significantly higher effectiveness of treatment than did the medium ( p.05) and the long groups ( p.005). Beginning treatment within the first 20 days was associated with a significantly high probability of excellent therapeutic response (OR 6.11; 95% confidence interval [CI], ), and beginning later was associated with a similar risk of poor response (OR 5.18; 95% CI, ). On the other hand, early intervention was associated with a five times greater risk of dropout than that of patients with delayed start of treatment (OR 4.99; 95% CI, ). The three subgroups were significantly ( p.05) different regarding the percentage of low and high responders. Conclusion: Our results showed a strong association between and functional outcome. Key Words: Italy; Rehabilitation; Stroke; Treatment outcome by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation From the I.R.C.C.S. Santa Lucia (Paolucci, Antonucci, Grasso, Morelli, Troisi, Coiro, Bragoni), and the Department of Psychology, University of Rome (Antonucci), Rome, Italy. Submitted September 8, Accepted in revised form October 29, Presented in part at the 2nd World Congress on Neurorehabilitation, April 1999, Toronto, Canada, and at the 8th European Stroke Conference, April 1999, Venice, Italy. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. Reprint request to Stefano Paolucci, MD, I.R.C.C.S. Santa Lucia, Via Ardeatina 306, 00179, Rome, Italy /00/ $3.00/0 doi: /apmr SEVERAL STUDIES on stroke recovery have been conducted to evaluate the role of medical, personal, demographic, and neuropsychologic variables in functional outcome Age and degree of disability at admission have been identified as strong prognostic factors influencing rehabilitation programs and amount of recovery. The strong relationship between age and disablement is well established. 13 Nakayama and coworkers 5 reported that a 10-year increase in age was associated with a 7% decrease in gain on Barthel Index (BI) scores; Kalra 6 found that after rehabilitation treatment elderly patients had lower BI scores and a lower rate of discharge to home; and Jeffery and Good 14 emphasized the association between older age and poor outcome. Similarly, the strong relationship between initial and later disability is well known. Johnston and colleagues 7 found that the severity of functional limitation was a determinant of later functional outcome; Shah and associates 15 reported that initial disability, measured as BI, was a powerful predictor of discharge BI score; and Oczkowski and Barreca 16 found that the absolute admission functional independence measure score was the best predictor of outcome disability and discharge destination. It is crucial to determine when to begin rehabilitation because best results usually occur within the first few months after stroke, 17,18 although some recovery can occur later. A short onset-to-admission interval (), with rehabilitation beginning quickly thereafter, has been recognized as a relevant favorable prognostic factor, 1,3,8,12,15,19 even though some authors have recently reported a favorable role for rehabilitation in chronic stroke patients However, is quite variable, depending on the clinical course of the acute phase and the number of beds available in rehabilitation wards. In some cases, admission to a rehabilitation ward is delayed because of the patient s comorbidities, and rehabilitation treatment can be started only after the stabilization of the patient s medical conditions. However, the relationship between an early start of rehabilitation and a better outcome was not observed by Johnston and coworkers, 7,23 who found earliness to be confounded with case severity. Moreover, multivariate models used in most outcome studies tend to be specific, but less sensitive, and do not allow a careful evaluation of the specific role of in functional outcome. The aim of this study was to evaluate the specific influence of on rehabilitation results in consecutive stroke inpatients, after age and disability matching to rule out the influence of recognized strong prognostic factors. In particular, the study compares rehabilitation results in homogeneous subgroups of patients admitted for rehabilitation of first stroke sequelae, and divided according to the period when they started specific rehabilitation. All subgroups began treatment within 60 days after stroke because this period is widely considered the optimal time for intensive rehabilitation. METHODS Subject Selection We evaluated stroke inpatient survivors admitted to our rehabilitation unit (64 beds) between September 1994 and

2 696 EARLY VS DELAYED INPATIENT STROKE REHABILITATION, Paolucci October 1998 for rehabilitation of sequelae of their first stroke. Rehabilitation staff included physicians (physiatrists, neurologists, urologists, and otorhinolaryngologists), neuropsychologists, nurses, physiotherapists, occupational and speech therapists, a social services care manager, dietitians, and support staff. Our rehabilitation unit is included in a large, freestanding, university-affiliated rehabilitation institute independent from the acute general hospital. Admission is possible for all recent stroke survivors with functional disability or cognitive loss (International Classification of Diseases, 9th Revision [ICD-9] diagnosis code 438), without medical conditions severe enough to contraindicate physical therapy, after a specific request by the acute general hospital (mainly the university neurologic clinic). In particular, the threshold criterion for hospital admission is the possibility to participate actively in rehabilitation and to tolerate daily intense treatment. In all patients, the rehabilitation unit s physicians perform a careful evaluation of medical conditions before admission. Unfortunately, availability of beds is generally lower than demand, and there is usually a waiting list for the hospital that is controlled by the hospital administration, according to a list reservation priority, in which ward physicians (including the researchers) are not involved. So, because of the separation of medical and administrative staff, researchers have no way to predict patients allocation and, consequently, no possibility to choose patients. As thoroughly detailed in the Results section, some patients were discharged home before hospital rehabilitation admission, and most of them underwent domiciliary conventional exercises while they were waiting for hospital admission. Stroke has been defined as a sudden, nonconvulsive, focal neurologic deficit persisting for more than 24 hours. 24 The diagnosis of stroke was based on history, clinical examination, and neuroradiologic findings (computed tomography [CT] or magnetic resonance imaging [MRI]). Exclusion criteria included subarachnoid hemorrhage and presence of sequelae of previous cerebrovascular accidents or of other chronic disabling pathologic conditions (ie, severe Parkinson s disease, polyneuropathy, severe cardiac, liver, or renal failure, cancer, limb amputation). We excluded patients who had negative CT scans or MRI in the subacute phase also, to avoid enrolling patients with transient ischemic attacks and to reduce the impact of spontaneous recovery. Neurologic and Functional Assessment Patients were assessed comprehensively by members of the multidisciplinary team on arrival. In particular, at admission, all patients underwent clinical, neurologic, neuropsychologic, and functional examinations. Severity of stroke and abilities in activities of daily living (ADL) were assessed on the Canadian Neurological Scale (CNS), with a cutoff score of 11.5 for normal patients, and on BI, with a cutoff score of 100 for independent patients. 25,26 The assessment of ADL was repeated after 2 days of stay, to rule out any cases of impaired functional status caused by the new hospitalization, and lastly at discharge. For statistical analyses, we used the second and third BI evaluations. At discharge, the CNS was not used because the target of the rehabilitation was functional and not neurologic recovery. Regarding neuropsychologic examination, aphasic patients were distinguished as Broca s, Wernicke s, or global aphasic according to their clinical evaluation and score on the Western Aphasia Battery. 27 Patients were also screened for neglect and classified as having neglect when they scored below the cutoff in three of four tests of a specific standardized battery, 28 including the Letter Cancellation Test, the Barrage Test, the Sentence Reading Test, and the Wundt- Jastrow AreaIllusion Test. Each patient (during acute hospital or rehabilitation stay) underwent at least a CT or an MRI brain examination. All CT or MRI examinations performed in the first 48 hours after the acute event were repeated in the subacute phase. To control for the prominent prognostic role of age and disability, patients were enrolled in homogeneous triads, matched for age (within 1 year) and BI score at admission, and different for to the rehabilitation ward. In particular, the matching for disability status included not only the same total BI score but also the same subscores. Each triad consisted of one patient with short (admitted within 20 days of the acute event), one patient with medium (admission ranging from days 21 to 40), and one patient with long (admission ranging from days 41 to 60). The threshold value of 20 days after stroke as the cutoff point for early intervention was chosen according to the mean time interval before admission to our hospital, as already published. 29 On the other hand, this value was intermediate between that of Rossi and coworkers 19 (day 12 since stroke) and that of Miyai and colleagues 22 (day 90). The evaluation of BI at admission and discharge was always performed by the same two physiatrists (one of the authors, DM, and a physiatrist on the ward, FR), who were not informed of the time of the acute event. Cohen s kappa measure of agreement between the two observers was.96 ( p.001). The rehabilitation plan, essentially based on practical ADL skills, was designed by the same physiatrist (DM) for all patients. Individual physiotherapy was performed for 60min twice a day (only once on Saturday), 6 days a week. All rehabilitation treatment began within 24 hours of admission, and each triad of patients was treated by the same therapists. If necessary, patients had access to individual training for neglect or speech therapy, 30,31 and for swallowing, bowel, and bladder dysfunctions. Physiotherapy and language treatment continued throughout the hospital stay, and the training for neglect lasted for 8 consecutive weeks. We calculated rehabilitation results using efficiency and effectiveness of treatment and percentage of low- and highresponse patients. 15,12 Efficiency is the amount of improvement in the rating score of each scale divided by length of rehabilitation stay; it represents the average increase per day obtained during rehabilitation stay. 15 Effectiveness reflects the proportion of potential improvement achieved during rehabilitation, calculated by the following formula: Effectiveness (Discharge score Initial score) (Maximum score Initial score) Therefore, if a patient obtains the top score after rehabilitation, effectiveness is 100%. As previously reported, 12 patients whose treatment effectiveness on ADL was lower than the mean 1 SD were considered as low-response patients. The model was studied according to the concept that in normal distribution, mean 1 SD generally includes two thirds of all observations. In the final sample, distribution of effectiveness on ADL was normal. In particular, for effectiveness on BI, the value of the ratio of skewness to its standard error value was 1.77 (.39/.22). Data Analysis and Statistics Baseline variables including sex; side of lesion; vocational status; educational status; type, side, and site of cerebral lesions; and presence of cognitive impairment (hemineglect, Broca s, Wernicke s, and global aphasia), depression, and comorbidities (heart disease, hypertension, and diabetes) were compared among the three groups by means of the likelihood ratio (LR) or analysis of variance. A first multiple logistic regression (forward stepwise, with.05 as probability for stepwise at entry) was performed in the whole sample using

3 EARLY VS DELAYED INPATIENT STROKE REHABILITATION, Paolucci 697 dropping out (coded as 1 present, 0 absent) as the dependent variable. The following dichotomous variables were used as independent variables (all coded as 1 or 0, depending on the presence or absence of the event, if not differently specified in parentheses): age ( 65yrs vs 65yrs), sex (coded as 1 male, 2 female), etiology of stroke (coded as 1 ischemic, 2 hemorrhagic), severity of stroke (CNS score 6 vs 6), (coded as 1 20 days from acute event, 2 20 days), and presence of hypertension, diabetes, heart disease, poststroke seizure, hemineglect, Broca s, Wernicke s, and global aphasia. Post hoc comparisons (Tukey High Significant Difference [HSD] test) were performed for efficiency, effectiveness, and length of stay of the three subgroups. Two multiple logistic regressions (forward stepwise, with.05 as probability for stepwise at entry) were performed using both low and high response on BI (both coded as 1 present, 0 absent) as the dependent variable. All independent variables were dichotomous: age ( 65yrs vs 65yrs), sex, etiology of stroke (ischemic or hemorrhagic), side of motor deficit (coded as 1 right hemiparesis/hemiplegia, 2 left hemiparesis/hemiplegia), severity of stroke (CNS score 6 vs 6),, and presence of poststroke seizures, hemineglect, Broca s, Wernicke s, and global aphasia. The reasons for the difference in the list of independent variables among the first and the latter two analyses are clinical: in the model predicting dropping out, we chose to evaluate the impact of medical variables, and in analyses studying functional outcome, we chose to evaluate the role of neurologic ones. RESULTS The final sample included 135 patients, divided into three age- and disability-matched subgroups of 45 patients. In all cases of patients in medium or long groups, delay in admission was caused by organizational problems (waiting list because a bed was not immediately available in the rehabilitation ward) and not by medical factors. During their stay in the acute care hospital, all patients were treated daily by physiotherapists to avoid secondary complications of contracture or pressure sores. Thirty patients of medium (66.7%) and all long patients were discharged home from the acute care hospital and then admitted to the rehabilitation hospital. Sixtytwo percent of the patients discharged home underwent domiciliary conventional exercises two or three times (40min/ session) a week while they were waiting for rehabilitation admission. None of these patients underwent speech therapy or cognitive training for neglect before admission to the rehabilitation hospital. Table 1 presents demographic, medical, neurologic, neuroradiologic, and functional findings of the three subgroups. No significant difference was found among the three subgroups for any parameter. In particular, severity of stroke (CNS score) was similar among subgroups, as was type, side, and site of cerebral lesions. Twelve patients (8.9%) did not complete treatment because of death or emergency transfer. As shown in figure 1, the three subgroups significantly differed (LR 7.23, df 2, p.05) in percentage of dropouts: we observed 17.8% of dropouts (4 deaths, 4 emergency transfers) in the short subgroup versus 6.67% in the medium (3 deaths) and 2.22% (1 transfer for renal failure) in the long subgroups. In particular, in the short 75% of both deaths and transfers were caused by cardiac complications; the remaining death was caused by a pulmonary embolism, and the remaining transfer was for treatment of a fracture of the paretic leg caused by an accidental fall. Regarding the three deaths in the medium Table 1: Demographic and Medical Characteristics of the Final Sample Short (n 45) Medium (n 45) Long (n 45) p BI score at admission NA Age (yrs) NS CNS score NS (days) NA Males 46.7% 44.4% 62.2% NS Right motor weakness 55.6% 46.7% 66.7% NS Schooling (yrs) NS MCA ischemic lesions 40.0% 42.2% 40.0% NS Lacunar lesions 17.8% 15.6% 13.3% NS VB ischemic lesions 8.9% 11.1% 8.9% NS Border or uncertain areas 17.8% 17.8% 17.8% NS Lobar hemorrhages 2.2% 2.2% 4.4% NS Putaminal hemorrhages 8.9% 4.4% 6.7% NS Thalamic hemorrhages 4.4% 6.7% 8.9% NS Broca s aphasia 8.9% 6.7% 2.2% NS Global aphasia 17.8% 20.0% 28.9% NS Hemineglect 17.8% 28.9% 17.8% NS Hypertension 31.1% 22.2% 31.1% NS Heart disease 48.9% 46.7% 48.9% NS Diabetes 13.3% 24.4% 11.1% NS Depression 31.1% 22.2% 31.1% NS Length of stay (days) NS Abbreviations: NS, not significant; NA, not applicable; BI, Barthel Index;, onset admission interval; CNS, Canadian Neurological Scale; MCA, middle cerebral artery; VB, vertebrobasilar. subgroup, the first was caused by a new stroke, the second by pulmonary embolism, and the third by acute pulmonary edema. In logistic regression with dropping out as the dependent variable, was the only variable to significantly enter the equation. The significance of the model was assessed by 2, which was (df 1, p.001). The final model yielded a 91.11% accuracy in prediction of the event. In particular, beginning treatment within the first 20 days from the acute event was associated with a five times greater risk of dropping out than of delaying start of treatment (OR 4.99; 95% CI, ). The mean BI score of the whole sample at discharge was , global effectiveness on BI was , and efficiency was Twenty-one (17.1%) and 29 (23.6%) of 123 patients who finished the study showed poor or excellent functional outcome, respectively. As shown in figure 2, the three subgroups significantly Fig 1. Percentages of dropouts, by onset-to-admission interval () subgroup (LR 7.23, df 2, p F.05).

4 698 EARLY VS DELAYED INPATIENT STROKE REHABILITATION, Paolucci Table 3: Results of Forward Stepwise Logistic Regression With Low Response on BI Score as Dependent Variable Independent Variables Dependent Variable: Low Response on BI B SE p OR 95% CI Severity of stroke (CNS score 6) Hemineglect days Constant Fig 2. Effectiveness on Barthel Index, by onset-to-admission interval () subgroup (*p F.05, short vs medium ; p F.005 short vs long ; Tukey HSD test). Values are mean standard error. Table 2: Results of Forward Stepwise Logistic Regression With High Response on BI Score as Dependent Variable Independent Variables Dependent Variable: High Response on BI B SE p OR 95% CI 20 days Age 65yrs CNS score Constant Only significant independent variables are shown. Significance of model: , df 3, p.001, accuracy in prediction 86.99%. Abbreviations: BI, Barthel Index; B, regression coefficient; SE, standard error; OR, odds ratio; CI, confidence interval;, onset admission interval; CNS, Canadian Neurological Scale. Only significant independent variables are shown. Significance of model: , df 2, p.001, accuracy in prediction 85.37%. Abbreviations: BI, Barthel Index; B, regression coefficient; SE, standard error; OR, odds ratio; CI, confidence interval;, onset admission interval; CNS, Canadian Neurological Scale. differed (F(2, 120) 6.21, p.005) in effectiveness of treatment, evaluated on BI. In particular, post hoc comparison (Tukey HSD test) showed that the short subgroup had a significantly higher treatment effectiveness than that of the medium ( p.05) and the long subgroups ( p.005). No significant difference in effectiveness was observed between the medium and the long subgroups. Rehabilitation efficiency on BI showed progressive worsening according to increasing, passing from (short ) to (medium ) and (long ). However, this trend was not significant, and it is presumably caused by increasing intragroup variability based on increasing. Table 2 shows the results of multiple logistic regression with occurrence of high response on BI as the dependent variable. In this model, of 20 days or less, age younger than 65yrs, and severity of stroke (CNS score at admission 6) were significantly associated with high therapeutic response. The significance of the model was p.001 ( , df 3), with 86.99% accuracy in prediction of therapeutic response. In particular, starting treatment within the first 20 days is associated with a six times greater probability of high response than delaying treatment. Table 3 shows the results of multiple logistic regression with low response on BI as the dependent variable. In this model, severity of stroke, hemineglect, and of more than 20 days were significantly associated with low therapeutic response. The significance of the model was p.001 ( , df 2), with 85.37% accuracy in prediction. In particular, starting treatment after day 20 is associated with a five times greater risk of low response than initiating early treatment. As presented in table 4, the three subgroups showed significant differences (LR 6.49, df 2, p.05 for low response; and LR 8.44, df 2, p.05 per high response, respectively) regarding the percentage of patients with poor or excellent therapeutic responses. In particular, a low therapeutic response was found in only 5.4% of the patients with early treatment as compared with 19% and 25% of the patients treated later. Also, a high response was found in 40.5% of the patients with early treatment versus 19% and 13.6% of the patients who began treatment later. DISCUSSION Our results underscore the importance of timing as a specific prognostic factor in rehabilitation results and confirm that early specific rehabilitation treatment is associated with greater improvement in ADL than delayed treatment. The best functional recovery occurs during the early weeks of treatment after the event, and effectiveness of stroke rehabilitation gradually decreases after the first weeks of treatment. In our series, the decreased effectiveness in the medium and the long subgroups was presumably related to delayed start of specific rehabilitation treatment. In fact, the three subgroups of patients differed only in delays in receiving specific rehabilitation. The patients in the three subgroups were not only matched for age and disability, but were also homogeneous for medical, neurologic, and neuroradiologic findings; also, the same inpatient rehabilitation treatment was carried out by the same therapists for all three subgroups. Although most of patients underwent a low-intensity rehabilitation program while they were waiting for rehabilitation admission, no patient received a specific multidisciplinary rehabilitation treatment, including speech therapy or cognitive training for neglect, before admission to the rehabilitation hospital. Therefore, the different rehabilitation results were probably due to both the more efficacious action and different therapeutic plans of specific multidisciplinary treatment performed in the rehabilitation ward on very recent stroke sequelae. The goals of inpatient specific treatment were to reduce disabilities in ADL and cognitive impairment, using techniques not available at home. Table 4: Frequency of Low and High Responders in the 3 Subgroups Short Medium Long Low responder on BI 5.4%* 19% 25% High responder on BI 40.5% 19% 13.6% Abbreviations: BI, Barthel Index;, onset admission interval. *LR 6.49, df 2, p.05. LR 8.44, df 2, p.05.

5 EARLY VS DELAYED INPATIENT STROKE REHABILITATION, Paolucci 699 As previously reported, cognitive training for hemineglect is effective not only in improving specific cognitive tasks, but also in influencing global functional outcome. 32 Moreover, intensity of treatment is clearly greater in the hospital ward than at home. Our data are consistent with the recent report of Rossi and coworkers, 19 who showed that patients with very early intervention after acute event have better outcomes. Therefore, even if most neurologic recovery occurs in the first 3 months after stroke 17 (and there is general agreement that this period is the optimal one for intensive rehabilitation), our data showed that earliness of rehabilitation itself seems to be a relevant prognostic factor of functional outcome, and it is important to begin treatment as soon as possible because any delay may greatly influence functional recovery. In our study, day 20 from the event was a threshold value for a different result of stroke rehabilitation: the behavior of the medium subgroup was more similar to that of the long rather than of the short, both for value of effectiveness and for percentage of low- or high-responder patients and for frequencies of dropouts. In fact, an early treatment greatly increases the probability of high therapeutic response, and conversely, a delay starting treatment increases the risk of low response. In particular, a treatment begun within the first 20 days was associated with a six times greater probability of high response compared with delayed treatment, and a treatment begun later with a five times greater probability of low response. In our previous published data, early rehabilitation within the first month from stroke was associated with nearly a twice greater risk of high response on BI than that of other patients. 12 So, the probability of high response increases according to the shortening of. In our series, delay in admission was caused by nonmedical factors: too few beds were available in the rehabilitation unit to accommodate the demand from the acute care hospital. Therefore, some patients waited in the general hospital, and during this period they did not have rehabilitation treatment or at most had conventional and unspecific treatment, and others were discharged to home before admission to the rehabilitation hospital. On the other hand, early admission to the rehabilitation hospital was associated with a five times greater risk of dropout than delayed start of treatment. However, the frequency of dropouts in our sample (8.9%) was lower than those reported by Siegler and coworkers 33 (14.7%) and by Wright and colleagues 34 (11.6%), and this discrepancy probably resulted from both the difference in subject selection (all strokes in our series vs a broad representation of several diagnostic categories in the latter two studies) and the careful clinical evaluation performed in our center before admission. The necessity of beginning rehabilitation treatment as soon as possible to obtain better rehabilitation results implies a potential risk of clinical emergencies, because in some cases the medical conditions in the first days after stroke are not yet stabilized. Even if these medical complications increase the costs of rehabilitation and cause distress to patients and their relatives, in our opinion a more favorable outcome of patients with short counterbalances the increased risk of dropouts. A careful evaluation of medical conditions before admission to a rehabilitation ward may reduce, but not eliminate the risk of new clinical events. In particular, we essentially observed cardiac complications. Cardiac complications are common during rehabilitation stay, 35 and are considered to be due more to excessive sympathoadrenal tone after stroke than to a concomitant ischemic cardiac disease. 36 According to this view, in our sample the presence of heart disease before stroke (similar among the three subgroups) did not emerge as a predictor of dropouts, confirming the hypothesis that cardiac complications after stroke cannot be explained only by previous cardiac disease; neurogenic cardiac effects of cerebral lesions also play a role. Stressful stimuli that might have been innocuous before stroke might cause cardiac damage because of increased sympathetic response. 36 Consistent with this hypothesis, the occurrence of cardiac complications (high in the first weeks) progressively decreased in the following weeks, together with the concomitant decrease of stressful stimuli. The lack of a control group without any treatment may reduce the power of our results. Spontaneous recovery of brain function after acute stroke might overlap with recovery attributable to rehabilitation. Available evidence suggests that stroke rehabilitation is effective, even if at present we cannot easily differentiate the influence of specific treatment and the natural recovery process. 37 To our knowledge, no studies of the course of poststroke recovery in the absence of any intervention have been done. Essentially, almost all studies of natural history occur in settings (such as nursing care) that actually include rehabilitative activities. 38 Moreover, it would be ethically and practically difficult to conduct a randomized trial. Therefore, earliness of rehabilitation itself seems to be a relevant prognostic factor of functional outcome, and it is important to begin treatment as soon as possible because any delay may greatly influence functional recovery, even though in some cases of early treatment the patient s clinical status may not yet be stabilized. References 1. Feigenson JS, McDowell FH, Meese P, McCarthy ML, Greensberg SD. Factors influencing outcome and length of stay in a stroke rehabilitation unit. Part 1. Analysis of 248 unscreened patients. Medical and functional prognostic indicators. Stroke 1977;8: Feigenson JS, McCarthy ML, Greensberg SD, Feigenson WD. Factors influencing outcome and length of stay in a stroke rehabilitation unit. Part 2. Comparison of 318 screened and 248 unscreened patients. Stroke 1977;8: Kotila M, Waltimo O, Niemi ML, Laaksonen R, Lempinen M. The profile of recovery from stroke and factors influencing outcome. Stroke 1984;15: Henley S, Pettit S, Todd-Pokropek A, Tupper A. Who goes home? Predictive factors in stroke recovery. J Neurol Neurosurg Psychiatry 1985;48: Nakayama H, Jorgensen HS, Raaschou HO, Olsen TS. The influence of age on stroke outcome. The Copenhagen Stroke Study. Stroke 1994;25: Kalra L. Does age affect benefits of stroke unit rehabilitation? Stroke 1994;25: Johnston MV, Kirshblum S, Zorowitz R, Shiflett SC. Prediction of outcomes following rehabilitation of stroke patients. Neurorehabilitation 1992;2: Shah S, Vanclay F, Cooper B. Predicting discharge status at commencement of stroke rehabilitation. Stroke 1989;20: Denes G, Semenza C, Stoppa E, Lis A. Unilateral spatial neglect and recovery from hemiplegia a follow up study. Brain 1982;105: Galski T, Bruno RL, Zorowitz R, Walker J. Predicting length of stay, functional outcome, and aftercare in the rehabilitation of stroke patients. The dominant role of higher-order cognition. Stroke 1993;24: Ween JE, Alexander MP, D Esposito M, Roberts M. Factors predictive of stroke outcome in a rehabilitation setting. Neurology 1996;47: Paolucci S, Antonucci G, Pratesi L, Traballesi M, Lubich S, Grasso MG. Functional outcome in stroke inpatient rehabilitation: predicting no, low and high response patients. Cerebrovasc Dis 1998;8:

6 700 EARLY VS DELAYED INPATIENT STROKE REHABILITATION, Paolucci 13. Chamie M. What does morbidity have to do with disability? Disabil Rehabil 1995;17: Jeffery DR, Good DC. Rehabilitation of the stroke patient. Curr Opin Neurol 1995;8: Shah S, Vanclay F, Cooper B. Efficiency, effectiveness and duration of stroke rehabilitation. Stroke 1990;21: Oczkowski WJ, Barreca S. The functional independence measure: its use to identify rehabilitation needs in stroke survivors. Arch Phys Med Rehabil 1993;74: Wade DT, Wood VA, Langton Hewer R. Recovery after stroke: the first three months. J Neurol Neurosurg Psychiatry 1985;48: Jørgensen HS, Nakayama H, Raaschou AO, Vive-Larsen J, Støier M, Olsen TM. Outcome and time course of recovery in stroke. Part II: time course of recovery. The Copenhagen Stroke Study. Arch Phys Med Rehabil 1995;76: Rossi PW, Forer S, Wiechers D. Effective rehabilitation for patients with stroke: analysis of entry, functional gain, and discharge to community. J Neurol Rehabil 1997;11: Dam M, Tonin P, Casson S, Ermani M, Pizzolato G, Battistin L. The effects of long-term rehabilitation therapy on poststroke hemiplegic patients. Stroke 1993;24: Tangeman PT, Banaitis DA, Williams AK. Rehabilitation of chronic stroke patients: changes in functional performance. Arch Phys Med Rehabil 1990;71: Miyai I, Suzuki T, Katsumasa K, Kang J, Kajiura I. Functional outcome of multidisciplinary rehabilitation outcome in chronic stroke. J Neurol Rehabil 1998;12: Johnston MV, Keiter M. Early rehabilitation for stroke patients: a new look. Arch Phys Med Rehabil 1984;65: Foulkes MA, Wolf PA, Price TR, Mohr JP, Hier DB. The stroke data bank: design, methods, and baseline characteristics. Stroke 1988;19: Coté R, Battista RN, Wolfson C, Boucher J, Adam J, Hachinski V. The Canadian Neurological Scale: validation and reliability assessment. Neurology 1989;39: Mahoney F, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J Rehabil 1965;14: Kertesz A. Aphasia and associated disorders: taxonomy, localization and recovery. New York: Grune and Stratton; Pizzamiglio L, Judica A, Razzano C, Zoccolotti P. Toward a comprehensive diagnosis of visual-spatial disorders in unilateral brain-damaged patients. Psychol Assess 1989;5: Paolucci S, Antonucci G, Pratesi L, Traballesi M, Grasso MG, Lubich S. Post-stroke depression and its role in rehabilitation of inpatients. Arch Phys Med Rehabil 1999;80: Pizzamiglio L, Antonucci G, Judica A, Montenero P, Razzano C, Zoccolotti P. Cognitive rehabilitation of the hemineglect disorder in chronic patients with unilateral right brain damage. J Clin Exp Neuropsychol 1992;14: Basso A. Approaches to neuropsychological rehabilitation: language disorders. In: Meter MJ, Benson AL, Dialer L, editors. Neuropsychological rehabilitation. Edinburgh: Churchill Livingstone; p Paolucci S, Antonucci G, Guariglia C, Magnotti L, Pizzamiglio L, Zoccolotti P. Facilitatory effect of neglect rehabilitation on the recovery of left hemiplegic stroke patients: a cross-over study. J Neurol 1996;243: Siegler EL, Stineman MG, Maislin G. Development of complications during rehabilitation. Arch Intern Med 1994;154: Wright RE, Rao N, Smith RM, Harvey RF. Risk factors for death and emergency transfer in acute and subacute inpatient rehabilitation. Arch Phys Med Rehabil 1996;77: Roth EJ. Heart disease in patients with stroke. Part II: impact and implications for rehabilitation. Arch Phys Med Rehabil 1994;75: Oppenheimer SM. Neurogenic cardiac effects of cerebrovascular disease. Curr Opin Neurol 1994;7: Gresham GE, Alexander D, Bishop DS, Giuliani C, Goldberg G, Holland A, et al. AHA Prevention Conference IV: rehabilitation panel. Stroke 1997;28: Wade D. Stroke. In: Greenwood R, Barnes MP, McMillan TM, Ward CD, editors. Neurological rehabilitation. Edinburgh: Churchill Livingstone; p