Descriptive characteristics as potential predictors of outcomes following constraint-induced movement therapy for people after stroke.

Limited evidence exists regarding the characteristics of people who achieve the best outcomes with constraint-induced movement therapy (CIMT). (1) Constraint-induced movement therapy is a rehabilitative strategy used primarily with the poststroke population to increase the functional use of the neurologically weaker upper extremity through massed practice while restraining the less-involved upper extremity. (2) Although stroke is the most common disabling condition in America, with 30% to 66% of people poststroke losing functional ability in their more-affected arm and hand, there are few research-supported interventions available to these individuals. (3,4)

Research evidence supports the effectiveness of CIMT, but many questions persist about who can benefit from this therapy. (1-3,5-10) Selection criteria for participation in CIMT should be carefully examined to determine who benefits most from this intervention and what characteristics are predictive of positive CIMT outcomes. The aim of this study was to establish a simple predictive model for CIMT outcomes based on 6 descriptive characteristics of people poststroke. The identification of clinical predictors for outcomes of CIMT is essential and of value to both researchers and clinicians. (11)

Method

Participants

A convenience sample of 55 participants was recruited from 2 local CIMT projects. Table 1 presents the main participant characteristics. Participants signed an informed consent statement prior to participation in the study. Inclusion and exclusion criteria are listed in Appendix 1. Components of the methods utilized in this study, including portions of the inclusion and exclusion criteria, were adapted from the EXCITE trial methods. (12)

Procedure

After a pretest, participants received 2 weeks (10 consecutive weekdays) of supervised task practice using their affected hand and arm for 6 hours per day. For this period, the unaffected hand was immobilized in a padded mitt for a goal of 90% of the participants' waking hours. The mitt was used at all times except when performing a minimal amount of agreed-on activities (eg, bathroom activities, when used for an assistive device in walking, when safety was compromised). An activity

log was kept by the trainer to chronicle what tasks had been attempted and how the tasks were progressed during training. The CIMT consisted of a set of tasks such as picking up pencils, moving beans from 1 container to another, and stacking. The treatment was focused on performance of frequent movement repetitions while performing functional activities. To remain challenging, as performance improved, tasks were increased in complexity and difficulty. Throughout the 2 weeks, the participants were strongly encouraged to continue to use their weaker hand during activities throughout the day and while at home. After the 6 hours of intensive therapy, participants returned home and maintained a diary documenting activities and mitt time use. During the weekends, there were no assigned tasks, but the participants were instructed to continue to wear their mitt and maintain a home diary. The 10 days of training were followed by an immediate posttest and a 4-to 6-month follow-up posttest. (1) The predictors at the follow-up posttest are of most interest, more specifically the predictors of what individuals will perform the best at 4 to 6 months after the treatment.

Outcome Measures

Two main outcome measures, commonly reported in CIMT studies, were used for this study: (1) the Wolf Motor Function Test (WMFT), a test of movement capability, and (2) the amount component of the MAL (MALa), a test of perceived use.

The WMFT is a commonly reported outcome measure in CIMT studies. (2,5,7,8) It evaluates movement capability through a series of 15 timed tasks and 2 strength tasks. Only the timed tasks were used in this study. The tasks progress from joint-specific movements to multijoint movements. (5,7) The reliability of WMFT scores has been reported, with interrater reliability (r) established at .93. (13) The WMFT outcome measurement is reported as a mean of the affected task times minus the mean of the unaffected task times.

The MAL, also a commonly used CIMT outcome measure, (2,5-8) is a 30-question structured interview in which the participants respond with a number corresponding to a given amount of use or perception of how well they have used their affected arm when away from the laboratory environment. For example, the participant would respond to the question: "How much do you use your more affected arm to turn on a light switch?" by choosing the appropriate response from the MALa (Appendix 2). Only the mean of the MALa was used as an outcome measure. The interrater reliability for the MAL is .94. (5,7)

Six Potential Predictors

Six prospective descriptive predictors were investigated: (1) side of stroke, (2) time since stroke, (3) hand dominance, (4) age, (5) sex, and (6) ambulatory status. These predictors were included in the regression model for several reasons. The main reasons for inclusion of these predictors is that many of them appear in stroke rehabilitation research as predictors of other outcomes, such as differing therapeutic interventions, return to function, and return to life roles. (14-24) Second, these predictors are often discussed in the CIMT literature as having a potential to affect outcomes. (3,5,7) Third, some predictors demonstrated strong predictive value in the pilot studies. The detailed reasons for selection of each predictor are elaborated below.

Side of stroke. Side of stroke is an obvious factor to consider when studying the effects of CIMT. A great deal of research has assessed right-brain versus left-brain functions, right-sided versus left-sided strokes, and the varying effects that these factors have on patients' presentation, functioning, and outcomes. Ornstein (25) stated that over the last 25 years more than 45,000 articles and books have been written on the 2 hemispheres. Although some CIMT studies have included participants with both left- and right-hemispheric damage, no main effect was found for side of hemiparesis. (7)

An individual who sustains a left-brain stroke may have an inability to solve problems, is often more easily angered and frustrated, has impaired retention of information, and may have language difficulties or apraxia. (26) Language difficulties may transfer into difficulty understanding directions for the therapy, or these individuals may be limited in keeping track of their home activity. With communication difficulty, there may be increased levels of frustration, especially given the social nature of CIMT that develops due to extensive time with trainers. Those with apraxia may have spatial or timing errors, such as a delay in initiation or inappropriate pauses. (14) These types of errors may impede acquisition (14) of new motor skills and thus result in increased difficulty with CIMT.

People with right-brain stroke, however, also have distinctive problems that could limit success with CIMT. They have left-side neglect, often have difficulty with spatial-perceptual tasks, are more impulsive, and frequently have greater balance problems. (26) Individuals with neglect may have significant difficulties attending to the CIMT tasks. Neglect would then result in interference with motor acquisition, leading to poorer outcomes. Conversely, they may be able to overcome the inattention caused by neglect, due to the constant attention-driven methods of CIMT, and demonstrate improvement. The various presentations of pathology attributed to side of lesion could affect outcomes following CIMT; therefore, it was included as a potential predictor.

Time since stroke. Constraint-induced movement therapy is based on the assumption that the nervous system always remains plastic; thus, the time of administration of therapy would not be a factor. Improvements then would be possible at any time following the insult. (2) Although people who are as much as 18 years poststroke have demonstrated functional improvements following CIMT, (3) the literature is not clear regarding whether individuals in the chronic phase of stroke can improve to the same extent as those who have sustained a stroke more recently. Time since stroke is often used as a predictor for stroke outcomes with other interventions. (15,17,27) Although the greater the time since stroke is associated with poorer functional recovery, the attention-driven methods of CIMT may counteract this temporal relationship, resulting in improvements at any time poststroke.

Hand dominance. Hand dominance affects an individual's functional ability after a stroke, yet current CIMT studies have not sufficiently addressed dominance and its role in functional recovery. (28) Hand dominance is a behavioral manifestation of hemispheric asymmetry. (28) One may consider that people with dominant-side hemiparesis would be more motivated to regain function of that extremity because it is the extremity of functional preference and practiced motor skill. Moreover, the dominant hemisphere has more intricate motor programs, with better-developed coordination and skill. Does this premorbid increased representation of the dominant hand remain poststroke, allowing for increased ease in recovery of function? In contrast, an individual with a nondominant-side stroke could possibly return to a previous level of functioning at a quicker pace compared with an individual with a dominant-side stroke. This possibility exists because of the decreased functional demands on the nondominant extremity. Dominance is an important issue that needs to be investigated as a predictor of function following CIMT.

Age. The research literature to date presents conflicting results regarding the role of age in rehabilitation. Motor performance factors have been demonstrated to be greatly influenced by age. (18,19,29) According to Jongbloed, (30) age was identified as a significant predictor in numerous studies. Specifically, these studies indicated that age is negatively correlated with functional return. Conflicting studies, however, have shown that age does not have a negative effect on function over time. (31) Research has demonstrated the benefits of intensive stroke rehabilitation programs regardless of age. (19) Kugler et al (18) stated that age should not be a limiting factor in the early rehabilitation of patients following a stroke. The inclusion of age in the regression equation as a potential predictor is important to help clarify the role that age plays in response to CIMT.

Sex. Scarce and conflicting information exists about sex differences in the management of people with stroke. (20) Research on sex differences following stroke is needed in order to provide useful insight for long-term intervention and establishment of appropriate therapy. (20) Wyller et al (21) studied sex differences in functional outcomes following stroke. Utilizing age-adjusted odds, they concluded that women seem to be functionally more impaired by stroke than men. Conversely, Twigg et al (22) found no significant correlation between sex and functional outcomes following stroke.

Ambulatory status. Ambulatory status was included as a potential predictor of upper-extremity recovery for several reasons. This predictor was included in the regression model because cumulative deficits poststroke can affect individuals' functional outcomes. (23) People who are nonambulatory, therefore, may have poorer outcomes. Moreover, although the most rapid recovery for both upper and lower extremities occurs within the first 30 days after stroke, the severity of motor impairments and the subsequent patterns of recovery are similar for both extremities. (32) A relationship exists, therefore, between upper- and lower-extremity motor recovery. (32)

In addition, the ability to walk is a strong predictor of functional outcomes (32) and has been included in regression models previously for stroke recovery. (24) People who are able to ambulate may be able to use their hand in more functional tasks. For the purposes of this study, ambulation was used as a dichotomous variable. If the participants were ambulating, with or without an assistive device, when seen at the clinic, then they were considered functionally ambulatory. If they were using a wheelchair most of the time, they were categorized as nonambulatory.

Data Analysis

For the intention-to-treat analysis, demographic and clinical characteristics were compared to determine differences that could result in a bias. These comparisons were made between those individuals who completed the follow-up posttest and those who did not. The continuous variables from the intention-to-treat analysis were analyzed using t tests, and categorical variables were analyzed using chi-square tests or Fisher exact tests. All data were analyzed using an intention-to-treat approach in which the pretest scores were used as the follow-up posttest scores for those participants who did not return for the follow-up.

The normality of the data for the WMFT and the MALa, the dependent variables, was visually verified with probability plots and statistically verified with the Kolmogorov-Smirnov f test. Only the WMFT required transformation using the natural log [ln] to meet assumptions of normality.

The independent variables were used to develop a general linear model for the dependent variables. These analyses were performed for the follow-up posttest. The pretest scores for each of the dependent measures were used as covariates to statistically control for individual differences that existed before treatment. A forward stepwise procedure was used in which the variables were examined at each step for entry into or removal from the model. The least significant variables were removed from the model, based on their level of association with the dependent variables at each step. Adjusted [R.sup.2] values, P values, and 95% confidence intervals were calculated. Thorough regression diagnostics were run and jackknife residual analyses were performed to verify the basic assumptions. Presence of multicollinearity among predictor variables in the regression models was assessed using a variance inflation factor. (1)

Results

Power Analysis

A post hoc power analysis was conducted to determine the power for the sample size (N = 55). The effect sizes ([f.sup.2]) for the WMFT and the MALa follow-up posttest were determined from the data, and they were 2.58 and 0.89, respectively. Using these effect sizes, the sample size of 55 participants at [alpha] = .05, for 6 predictors, met an average power level of [beta] = 1.0. This strong power can be attributed primarily to the presence of the covariate in the model.

Intention-to-Treat Analysis

An intention-to-treat analysis was used because of the significant pretest differences apparent between the group that completed the follow-up posttest and the group that did not. Nine participants (16%) did not return for the follow-up posttest evaluation, and their pretest score was used as the follow-up posttest measure. (33) Participants who dropped out of the study showed significantly lower ability levels; therefore, an intention-to-treat analysis was essential to avoid bias in the results.

Multiple Regression Modeling

The potential predictors were entered into 2 multiple linear regression models with stepwise entry using the following dependent measures: (1) [ln] WMFT follow-up posttest and (2) MALa follow-up posttest. None of the potential predictor variables made significant contributions to the WMFT model. The B and P values for all the independent variables are listed in Table 2.

At the follow-up posttest, age was the only significant predictor variable for the MALa model. This model accounted for 0.472 of the variance in MALa scores at the follow-up posttest. To aid in understanding, a graphical example of the MALa is presented in the Figure. This Figure is meant as an example to demonstrate and visually explain the interpretation of the regression equation. The regression equation is:

MALafu' = 2.308 + (1.047) (MALapre)

- (0.0196) (Age)

where MALafu' is the predicted MALa follow-up posttest score and MALapre is the MALa pretest score.

[FIGURE OMITTED]

The largest value of the variance inflation factors was <1.07, indicating that multicollinearity among the predictors did not unduly influence the regression estimates. (34) Residuals appeared to be normally distributed, and presence of outliers was assessed using jackknife residuals. The sample contained 2 outliers for the WMFT model. The influence and accuracy of these data points were assessed, and they remained in the model. (35) These post hoc regression diagnostics results suggested that the regression analysis was appropriate.

Discussion

Although CIMT has been proven to be a beneficial therapy for people poststroke, (2,3,5-10) little is known about the characteristics that are predictive of CIMT outcomes. (1) The objective of this study was to establish a usable model, not an extensive or comprehensive models. Although detailed models serve a function, these models are less readily utilized and often have little predictive capability over simple models. (36,37) Other predictive variables could have been included in this model, potentially making it more inclusive. For example, memory retention, mood, or motivational variables may have increased the predictive capability of this model. In addition, lesion location, size, or type could have been assessed using imaging techniques and included in the model because these variables are known to be valuable in prediction of stroke outcomes. (38) These additional predictors, however, would add to the complexity of the model, making it more difficult to use and clinically less practical.

The goal of this study was to determine the significance of 6 descriptive predictors of outcomes following CIMT. Age was the only descriptive characteristic that demonstrated a predictive relationship. This relationship was with the MALa outcome, a test for the participants' perception of how much they use their more affected arm. None of the potential predictors showed a predictive relationship with the WMFT. This may be the most significant finding because it emphasizes the importance of not excluding individuals based on time since stroke or other included descriptive predictors used in this model. These descriptors do not predict an individual's outcomes following CIMT.

Age was the only independent variable that had a predictive relationship with amount of use of the more affected upper extremity following CIMT. Age is only a predictor for the amount section of the MAL at follow-up posttest. Assuming the same baseline MALa score, as age increases, the predicted MALa score at follow-up posttest decreases. There is an inverse relationship, therefore, between MALa score and age. Although there is a general belief that younger people have greater potential for recovery, (30,31) research has demonstrated the benefits of intensive stroke rehabilitation programs regardless of age. (19) In our study, age was only a predictor of long-term amount of use of the more affected upper extremity, not movement capability of the affected upper extremity, as measured by the WMFT.

The most ideal candidates, those who will maintain improvement the longest, are those who are younger. If clinicians, therefore, only want to offer traditional CIMT to people who will maintain the most gains in amount of use of the affected upper extremity in the long-term, the results of this study suggest offering it to people who are younger. Important to clarify, however, is the fact that older individuals make gains in amount of use with therapy; they just do not seem to maintain these gains as long as younger individuals. In a previous study, (1) significant predictive ability was discovered with finger extension and grasp release for a predictor of long-term improvement on the WMFT. Therefore, when age and finger extension are used in their appropriate regression equations, along with the pretest scores, the therapist can predict an individual's score on the MALa or WMFT, respectively. These findings can be used to help determine the appropriate individuals for participation in traditional CIMT.

Side of stroke, time since stroke, hand dominance, sex, and ambulatory status were not predictive of the outcome measures following CIMT during the follow-up time frame. Knowing what variables are not predictive is as valuable and knowing the predictors. For example, when screening a potential client for CIMT, individuals should not be discriminated against for participation in CIMT based on side of stroke, chronicity, hand dominance, sex, or ambulatory status. This result should be encouraging for those individuals with chronic disability for stroke, because time since stroke was not a predictor of outcomes in this study. The individual with the most acute stroke in this sample was approximately 7 months poststroke. A sample that includes people with acute stroke, however, may have different results.

The role that hand dominance plays in CIMT and stroke rehabilitation is a frequently debated and questioned topic. Intuitively, one may assume that people with dominant-side hemiparesis would be more motivated to regain function of the affected extremity. Our findings support those of Miltner et al, (7) which suggest that dominance does not seem to be a factor in outcomes for CIMT.

Of considerable interest is that none of the potential predictors emerged as significant predictors for the WMFF. In essence, this means that all people, defined across these 6 predictors, benefited equally in movement capability as a result of the intervention. This is a significant finding because it further emphasizes the importance of not excluding individuals based on time since stroke or other included descriptive predictors used in this model.

Limitations

Our sample was not randomly selected, but comprised respondents to inquiries for participants who met strict inclusion and exclusion criteria. Second, sample size decreased at the follow-up posttest evaluation, leading to a bias for withdrawals. Although this bias was dealt with during analysis, using an intention-to treat approach, its presence as a limitation is still significant. (1) These results should be replicated in another sample.

Including the pretest scores as a covariate in the model may be considered a limitation because it increases complexity and practicality of using the model. Specifically, a baseline test is required in order to utilize the regression equations for prediction of outcomes for CIMT. The use of the covariate, however, was essential to form accurate regression models. The pretest score for the WMFT was used as a covariate to statistically control for differences that existed among participants before treatment. (1)

Conclusions

Limited evidence exists regarding the specific descriptive characteristics of people who benefit most from CIMT. (1) Significant predictive ability was discovered with age of the participant. An inverse relationship was demonstrated between age and amount of use. When entering age in the appropriate regression equations, along with the covariate, an individual's CIMT outcome can be predicted. Although age was the only significant predictor, an equally strong finding in this study was that side of stroke, chronicity, hand dominance, sex, and ambulatory status were not found to be predictors at the follow-up session. This finding emphasizes the importance of not excluding people from CIMT based on these predictors.

Appendix 1.

Inclusion and Exclusion Criteria for Participants (a)

Inclusion

Diagnosis of at least 1 stroke and not >3 strokes on the same side of the brain

Stroke at least 6 months prior to study participation

Follow simple instructions (score of 20 or higher on Mini-Mental State Examination)

Sit independently without back or arm support for 5 minutes

Stand with support of a straight cane, quad cane, or hemiwalker for 2 minutes

Actively participate in 6 hours of therapy without long rest/nap periods

PROM of all UE motions of at least half the normal range

Exclusion

Health problems that put the participant at significant risk for harm during the study

Other neurological conditions

Medications for spasticity

Pain limiting participation in the study

(a) PROM=passive range of motion, UE=upper extremity.

Appendix 2.

Participant Response Choices for Motor Activity Log (MAL) Amount Scale

0.0 Did not use my weaker arm for the activity (not used)

0.5

1.0 Occasionally tried to use my weaker arm for that activity (very rarely)

1.5

2.0 Sometimes used my weaker arm for that activity but did most of the activity with my stronger arm (rarely)

2.5

3.0 Used my weaker arm for that activity about half as much as before the stroke (half prestroke)

3.5

4.0 Used my weaker arm for that activity almost as much as before the stroke (3/4 or 75% prestroke)

4.5

5.0 Used my weaker arm for that activity as much as before the stroke (same as prestroke)

This article was received September 2, 2005, and was accepted December 23, 2005.

References

(1) Fritz SL, Light K, Patterson T, et al. Active finger extension predicts outcomes after constraint-induced movement therapy for individuals with hemiparesis after stroke. Stroke. 2005;36:1172-1177.

(2) Morris D, Crago J, DeLuca S, et al. Constraint-induced (CI) movement therapy for motor recovery after stroke. Neurorehabilitation. 1997;9:29-43.

(3) Taub E, Uswatte G, Pidikiti R. Constraint-induced movement therapy: a new family of techniques with broad application to physical rehabilitation--a clinical review. J Rehabil Res Dev. 1999;36:237-251.

(4) Kwakkel G, Kollen B, Wagenaar R. Therapy impact on functional recovery in stroke rehabilitation: a critical review of the literature. Physiotherapy. 1999;13:457-470.

(5) Kunkel A, Kopp B, Muller G, et al. Constraint-induced movement therapy for motor recovery in chronic stroke patients. Arch Phys Med Rehabil. 1999;80:624-628.

(6) Blanton S, Wolf SL. An application of upper-extremity constraint-induced movement therapy in a patient with subacute stroke. Phys Ther. 1999; 79:847-853.

(7) Miltner W, Bauder H, Sommer M, et al. Effects of constraint-induced movement therapy on patients with chronic motor deficits after stroke: a replication. Stroke. 1999;30:586-592.

(8) Wolf SL, Blanton S, Baer H, et al. Repetitive task practice: a critical review of constraint-induced movement therapy in stroke. Neurology. 2002;8:325-338.

(9) Dromerick A, Edwards D, Hahn M. Does the application of constraint-induced movement therapy during acute rehabilitation reduce arm impairment after ischemic stroke? Stroke. 2000;31: 2984-2988.

(10) Taub E, Morris DM. Constraint-induced movement therapy to enhance recovery after stroke. Curt Atheroscler Rep. 2001;3:279-286.

(11) Hsieh CL, Sheu CF, Hsueh IP, Wang CH. Trunk control as an early predictor of comprehensive activities of daily living function in stroke patients. Stroke. 2002;33:2626-2630.

(12) Wolf SL. Extremity Constraint-Induced Therapy Evaluation (EXCITE). Bethesda, Md: National Institutes of Health; 2000. NCMRR-funded grant (R01HD037606-01A1).

(13) Morris DM, Uswatte G, Crago JE, et al. The reliability of the Wolf Motor Function Test for assessing upper extremity function after stroke. Arch Phys Med Rehabil. 2001;82:750-755.

(14) Heilman KM, Watson RT, Valenstein E. Neglect and Related Disorders: Clinical Neuropsychology. New York, NY: Oxford University Press; 1993: 279-386.

(15) Kwakkel G, Kollen BJ, van der Grond J, Prevo AJ. Probability of regaining dexterity in the flaccid upper limb: impact of severity of paresis and time since onset in acute stroke. Stroke. 2003;34:2181-2186.

(16) Nakaymna H. Recovery of upper extremity function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil. 1995;76:27-32.

(17) Duncan PW, Goldstein LB, Matchar D, et al. Measurement of motor recovery after stroke: outcome assessment and sample size requirements. Stroke. 1992;23:1084-1089.

(18) Kugler C, Altenhoner T, Lochner P, Ferbert A. Does age influence early recovery from ischemic stroke? A study from the Hessian Stroke Data Bank. J Neurol. 2003;250:676-681.

(19) Bagg S, Pombo AP, Hopman W. Effect of age on functional outcomes after stroke rehabilitation. Stroke. 2002;33:179-185.

(20) Di Carlo A, Lamassa M, Baldereschi M, et al. Sex differences in the clinical presentation, resource use, and 3-month outcome of acute stroke in Europe: data from a multicenter multinational hospital-based registry. Stroke. 2003;34:1114-1119.

(21) Wyller TB, Sodring KM, Sveen U, et al. Are there gender differences in functional outcome after stroke? Clin Rehabil. 1997;11:171-179.

(22) Twigg A, Cifu D, Keyser-Marcus L, Swartz Z. The association between gender, race and marital status on functional outcome at rehabilitation discharge after thromboembolic stroke: a prospective analysis. Neurorehabilitation. 1988;11:249-254.

(23) Patel AT, Duncan PW, Lai SM, Studenski S. The relation between impairments and functional outcomes poststroke. Arch Phys Med Rehabil. 2000;81:1357-1363.

(24) Counsell C, Dennis M, McDowall M, Warlow C. Predicting outcome after acute and subacute stroke: development and validation of new prognostic models. Stroke. 2002;33:1041-1047.

(25) Ornstein R. The Right Mind: Making Sense of the Hemispheres. New York, NY: Harcourt Brace & Co; 1997.

(26) Umphred D. Neurological Rehabilitation. St Louis, Mo: Mosby; 1995.

(27) Nakayama H, Jorgensen HS, Raaschou HO, Olsen TS. Recovery of upper extremity function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil. 1994;75:394-398.

(28) Hellige J. Hemispheric Asymmetry: What's Right and What's Left? Cambridge, Mass: Harvard University Press; 1993.

(29) Ward NS, Frackowiak RS. Age-related changes in the neural correlates of motor performance. Brain. 2003;126(pt 4):873-888.

(30) Jongbloed L. Prediction of function after stroke: a critical review. Stroke. 1986;17:765-776.

(31) Lehmann JF, DeLateur BJ, Fowler RS, et al. Stroke rehabilitation: outcome and prediction. Arch Phys Med Rehabil. 1975;56:383-389.

(32) Duncan PW, Goldstein LB, Homer RD, et al. Similar motor recovery of upper and lower extremities after stroke. Stroke. 1994;25:1181-1188.

(33) Hollis S, Campbell F. What is meant by intention to treat analysis? survey of published randomised controlled trials. BMJ. 1999;319 (7211) : 670-674.

(34) Neter J, Kutner M, Nachtsheim C, Wasserman W. Applied Linear Statistical Models. 4th ed. Chicago, Ill: McGraw-Hill College; 1996.

(35) Pigeot I. The jackknife and bootstrap in biomedical research: common principles and possible pitfalls. Drug Informational Journal. 2001;35:1431-1443.

(36) Gladman JR, Harwood DM, Barer DH. Predicting the outcome of acute stroke: prospective evaluation of five multivariate models and comparison with simple methods. J Neurol Neurosurg Psychiatry. 1992; 55:347-351.

(37) Stone SP, Allder SJ, Gladman JR. Predicting outcome in acute stroke. Br Med Bull. 2000;56:486-494.

(38) Baird AE, Dambrosia J, Janket S, et al. A three-item scale for the early prediction of stroke recovery. Lancet. 2001;357 (9274):2095-2099.

SL Fritz, PT, PhD, is Clinical Assistant Professor, Department of Exercise Science, University of South Carolina, 1300 Wheat St, Blatt PE Bldg, Columbia, SC 29208 (USA) ([e-mail]). The data were collected while Dr Fritz was a graduate student at Department of Physical Therapy, College of Public Health and Health Professions, University of Florida, Gainesville, Fla, and a predoctoral fellow at VA Brain Rehabilitation Research Center, Malcolm Randall VA Medical Center, Gainesville, Fla. Address all correspondence to Dr Fritz.

KE Light, PT, PhD, is Associate Professor, Department of Physical Therapy, College of Public Health and Health Professions, University of Florida.

SN Clifford, PT, MPT, is a graduate student at Department of Physical Therapy, University of Pittsburgh, Pittsburgh, Pa, and Assistant Professor, Department of Physical Therapy, Chatham College, Pittsburgh, Pa.

TS Patterson, MEd, is a graduate student in rehabilitation sciences at Department of Physical Therapy, College of Public Health and Health Professions, University of Florida.

AL Behrman, PT, PhD, is Associate Professor, Department of Physical Therapy, College of Public Health and Health Professions, University of Florida, and Research Investigator, VA Brain Rehabilitation Research Center, Malcolm Randall VA Medical Center, Gainesville, Fla.

SB Davis, PT, is Research Physical Therapist, VA Brain Rehabilitation Research Center, Malcolm Randall VA Medical Center.

Dr Fritz, Dr Light, and Dr Behrman provided concept/idea/research design and fund procurement. Dr Fritz and Dr Light provided writing. Dr Fritz, Ms Patterson, and Ms Davis provided data collection, and Dr Fritz and Ms Clifford provided data analysis. Dr Fritz, Dr Light, and Ms Patterson provided project management. Dr Fritz, Dr Light, and Ms Davis provided subjects. Dr Light and Ms Davis provided facilities/equipment. Dr Light and Dr Behrman provided institutional liaisons. Dr Light, Ms Clifford, and Ms Patterson provided consultation (including review of manuscript before submission). The authors are grateful to all of the therapists, physicians, and trainers who participated in subject recruitment, examination, and training. The cooperation of a large group of scientists made this project possible. A special thanks to Stephen Nadeau, MD, for his guidance and participation.

This study was approved by the Institutional Review Board of University of Florida.

This research, in part, was presented at the Combined Sections Meeting of the American Physical Therapy Association; February 23-27, 2005; New Orleans, La.

This study was supported, in part, by the Office of Research and Development Rehabilitation R&D Service, Brain Rehabilitation Research Center, Department of Veterans Affairs, Gainesville, Fla; Florida Biomedical Grant #BM042 (Light, principal investigator); and a Foundation for Physical Therapy Promotion of Doctoral Studies Scholarship grant.

Table 1.
Descriptive Characteristics of Participants

                            N     Min    Max

Time since stroke (d)       55    229    16,060 (45 y)
Age (y)                     55     22        83

                            Median   [bar.X]    SD

Time since stroke (d)       940      1,673      2,378
Age (y)                      66         62.05      14.6

                                               Frequency    %

Side of stroke         L brain                 28           51
                       R brain                 27           49

Dominance              Dominant side           26           47
                       Nondominant side        29           53

Sex                    Female                  22           40
                       Male                    33           60

Ambulatory status      Ambulator               44           80
                       Nonambulator            11           20

(a) Min=minimum, max=maximum, L=left, R=right.

Table 2.
Adjusted [R.sup.2] for Natural Logarithm [ln] Wolf Motor Function
Test (WMFT) Model and Motor Activity Log Amount Scale (MALa) Model (a)

Follow-up Posttest                  [In]WMFT

Adjusted [R.sup.2]                  0.721
F(sig)                              140.459 (.000)

Independent
Variables                           B            P (CI)

Covariate                            0.038       .000 (.032, .045)
Side of stroke (right=0, left=1)     0.140       .055
Time since stroke                    0.044       .552
Hand dominance (non=0, dom=1)        0.115       .110
Age                                 -0.089       .230
Sex (male=0, female=1)               0.071       .329
Ambulatory status (no=0, yes=1)      0.009       .907
Constant                             1.174       .000 (.819, 1.529)

Follow-up Posttest                  MALa

Adjusted [R.sup.2]                  0.472
F(sig)                              25.093 (.000)

Independent
Variables                           B            P (CI)

Covariate                            1.047       .000 (.717, 1.378)
Side of stroke (right=0, left=1)     0.018       .858
Time since stroke                    0.086       .400
Hand dominance (non=0, dom=1)        0.152       .127
Age                                 -0.020       .026 (-.037, -.002)
Sex (male=0, female=1)               0.047       .638
Ambulatory status (no=0, yes=1)      0.104       .350
Constant                             2.308       .000 (1.148, 3.467)

(a) B for weights, P for predictors and confidence intervals (CIs) for
the significant predictors. Non = nondominant, dom = dominant.