To the Editor:
Van der Lee and coworkers 1 reported that Constraint-Induced Movement (CI) therapy, compared with bimanual neurodevelopmental treatment (NDT), results in clinically insignificant improvements in the function of the more-impaired arm of persons with chronic stroke. This small effect stands in contrast to the very large improvements in real-world arm use obtained with CI therapy in 3 published studies2 3 4 (effect sizes 1.9 to 2.8, Motor Activity Log (MAL);2 5 large effect size 0.8),6 an as-yet-unpublished placebo-controlled study from our laboratory (effect size 2.5, MAL),7 8 2 conference presentations,9 10 and pilot studies from 6 laboratories involved in a national clinical trial of CI therapy with patients with subacute stroke (E. Taub, S. Wolf, C. Winstein, C. Giuliani, D. Good, K. Light, C. Kulkulka, D. Nichols, unpublished data, 1998). Van der Lee et al did not obtain a large treatment effect with their experimental group because the intensity of the therapy provided was inadequate and the subjects selected did not have a large enough deficit before treatment to show substantial benefits from therapy. In addition, we argue that the CI therapy group did not show significantly greater improvement than the bimanual NDT group because this group does not present an appropriate comparison.
In this laboratory, primary attention in CI therapy administration is given to training patients on a massed practice basis.2 7 8 Patients perform repetitive, behaviorally relevant arm movements with short intertrial and intertask rest periods for 6 to 7 hours daily for at least 2 consecutive weeks. Supervision is provided by a therapist on a continuous, one-on-one basis to ensure intensive practice.11 12 Van der Lee et al1 indicate that their experimental patients were treated in groups of 4 by 1 or 2 therapists and that tasks included housekeeping activities, handicrafts, and games. It is unlikely that this treatment format provided patients with adequately intensive practice. When van der Lee and a coauthor visited our laboratory for a week, they described their intended procedure in detail as the performance of hobby-type activities and some ADLs in a relaxed atmosphere without any systematic attempt to get patients to use the more-affected extremity intensively. The actual conduct of the therapy in this fashion was confirmed in a later letter from van der Lee. Before the investigators left this facility, they were strongly advised by 2 senior staff members (J. Crago, MSPT, and S. DeLuca, MA) that what they were planning to do would not work well because it was not sufficiently intense. Their protocol appeared so diffuse that they were told that it would test the minimum intensity of practice needed to produce a clinically meaningful effect.
The upper cutoff for experiment intake in our laboratory is a score of 2.5 (less than half as much use of the more-impaired extremity as before the stroke) on the Amount of Use (AOU) scale (range 0 to 5) of the Motor Activity Log (MAL), a semistructured interview of extremity use in the life situation.2 5 This cutoff is used because the brain injuries of stroke patients impose an upper physiological limit on the amount of improvement that can be produced (a score of 4 on the AOU scale represents “almost normal use”). Furthermore, patients with scores >2.5 do not suffer importantly from “learned nonuse,” which is defined as diminished extremity use in the real-world setting relative to ability as measured by a laboratory motor test. Learned nonuse is the target of CI therapy. The mean pretreatment score of experimental subjects in the study of van der Lee et al1 is 2.2 (SD=1.0), which is significantly greater than the scores (mean=1.1, SD=0.7, n=40) of patients in our experiments (P<0.001). The high score indicates that (1) a large minority of patients in the experimental group were too high functioning to have been included in any of the studies cited in the first paragraph and (2) there was little room for treatment-induced improvement in extremity use in the experimental group. In addition, the experimental subjects were higher functioning than their reference subjects on most measures, particularly the MAL, before treatment, thus decreasing the opportunity for improvement relative to the reference group.
In using a reference group given bimanual training based on NDT for 6 hours per day, van der Lee et al ignore our reports that conventional physical therapy, when administered in massed practice fashion, produces an effect that is as good as treatment involving training of the more-affected extremity and restraint of the less-affected extremity.5 7 8 13 These data, in combination with other findings,5 7 8 13 have identified the effective therapeutic factor in CI therapy as being the massing or concentration of practice, however achieved. It is primarily this factor that is thought to give rise to the massive increase in use-dependent cortical reorganization14 and other large changes in brain activity9 15 16 that appear to be the basis of the long-term changes in motor function reported by us.2 3 4 8 12 In this light, the (small) treatment effect for the bimanual NDT reference group is to be expected. There are no articles evaluating NDT on nearly as concentrated a basis as employed by van der Lee et al; in effect, they had 2 experimental groups receiving what could be considered 2 different forms of attenuated CI therapy.
Van der Lee et al1 also criticize the MAL on the grounds that it is a subjective measure and has no established validity. Although the MAL relies on self-report, it is a psychometrically robust instrument. The MAL is stable over a 2-week waiting period3 and over a 2-week placebo treatment period,7 8 13 and it has (1) high internal consistency (Cronbach’s α=0.88 to 0.95), (2) high interrater reliability (patient compared with primary caregiver, intraclass correlation type [3,1]17 =0.90),5 and (3) high test-retest reliability (r=0.94, P<0.01). Evidence of validity includes the 0.90 correlation between patient and caregiver reports, a perfect correlation with an observational measure of arm use (Actual Amount of Use Test),5 7 and a strong association between gains on the MAL and brain reorganization.9 14 15 16
The research from this laboratory was supported by grants B93–629AP and B95–975R from the Rehabilitation Research and Development Service, US Department of Veterans Affairs, and grant HD 34273 from the National Institutes of Health.
- Copyright © 2000 by American Heart Association
van der Lee JH, Wagenaar RC, Lankhorst GJ, Vogelaar TW, Deville WL, Bouter LM. Forced use of the upper extremity in chronic stroke patients: results from a single-blind randomized clinical trial. Stroke. 1999;30:2369–2375.
Miltner WHR, Bauder H, Sommer M, Dettmers C, Taub E. Effects of constraint-induced movement therapy on chronic stroke patients: a replication. Stroke. 1999;30:586–592.
Uswatte G, Taub E. Constraint-induced movement therapy. new approaches to outcome measurement in rehabilitation. In: Stuss DT, Winocur G, Robertson IH, eds. Cognitive Neurorehabilitation. Cambridge, UK: Cambridge University Press; 1999:215–229.
Cohen J. Statistical Power Analysis for the Behavioral Sciences. New York, NY: Academic Press; 1997.
Wittenberg GF, Chen R, Ishii K, Croarkin E, Eckloff S, Gerber LH, Taub E, Hallett M, Cohen LG. Task-related and resting regional cerebral blood flow changes after constraint-induced rehabilitation therapy. Paper scheduled for presentation at: Meeting of the American Academy of Neurology; May 2000; San Diego, Calif.
Koelbel S, Elbert T, Rockstroh B, Sterr A, Jahn T, Taub E. Constraint-induced movement therapy (CIMT) under German conditions. Paper presented at: Meeting of the International Psychophysiological Society; May 1997; Konstanz, Germany.
Morris DM, Crago JE, DeLuca SC, Pidikiti RD, Taub E. Constraint-induced (CI) movement therapy for motor recovery after stroke. Neurorehabilitation. 1997;9:29–43.
Taub E, Wolf SL. Constraint-induced (CI) movement techniques to facilitate upper extremity use in stroke patients. Top Stroke Rehabil. 1997;3:38–61.
Bauder H, Sommer M, Taub E, Miltner WHR. Effect of CI therapy on movement-related brain potentials. Psychophysiology. 1999;36:S31. Abstract.
We agree with Drs Taub and Uswatte that the results of the single-blind, randomized, controlled trial we publishedR1 seem to contradict the results of the studies published by Taub et al,R2 Miltner et al,R3 and Kunkel et al.R4 There are, however, important methodological differences between the studies mentioned by Taub and Uswatte and our study, and we consider this to be the most probable explanation of the apparent differences in results. Obviously, we are unable to comment on the unpublished studies mentioned by Taub and Uswatte. The studies presented by Kunkel et al and Miltner et al are uncontrolled case-series studies. Uncontrolled studies can be misleading and will almost always report grossly overestimated treatment effects.R5 There are 2 principal methodological differences between the randomized trial published by Taub et alR2 and our study: the number of patients (9 versus 66, respectively) and the intervention in the control group (“attention-control” versus intensive bimanual treatment, respectively). The contrast between the so-called “attention-control” intervention and the experimental intervention applied by Taub et al comprises not only specific but obviously also nonspecific aspects. In our opinion, this is illustrated by Figure 3 included in the article of Taub et al, in which half of the effect of the treatment in the experimental group appears to take place on day 1 of the treatment. In their letter, Taub and Uswatte mention 3 tentative explanations as to why we did not find a greater treatment effect: inadequate intensity of therapy, inadequate selection of patients, and inadequate contrast between the experimental and the control interventions. Furthermore, they contradict our criticism of the Motor Activity Log (MAL). We will reply to each of these items separately.
Intensity of treatment. We are somewhat puzzled by this criticism from Taub and Uswatte, because it is contradicted by their later statement that the improvement in our reference group was the result of the high intensity of treatment (“concentrated schedule of delivery”). Indeed, the intensity of treatment was high in both the experimental and the reference groups in our trial.
Selection of patients. It is true that we used a different outcome measure to the MAL to establish the upper cutoff for inclusion in our study, ie, the Action Research Arm (ARA) test, which was a primary outcome measure in our study. It is, indeed, conceivable that at higher pretreatment levels the MAL suffers from a ceiling effect. Nevertheless, we did find differential effects on the MAL, but these did not last.R1 To investigate the assumption that a more rigourous selection of patients would have yielded greater treatment effects, we reanalysed the subgroup of patients in our trial whose pretreatment MAL Amount Of Use (AOU) score was <2.5, as suggested by Taub and Uswatte. Somewhat to our surprise, however, we did not find a statistically significant difference in effectiveness in this subgroup (n=43), the mean difference in improvement on the MAL AOU scale between groups being 0.32 points (95% CI −0.10 to 0.75). This means that it could also be argued that higher-scoring patients are more liable to improve than patients with less residual arm function. On the ARA test, the difference in improvement between groups was significant in this subgroup, ie, 3.4 points (95% CI 1.3 to 5.6) and not notably different from the effect in the entire study population, ie, 3.0 points (95% CI 1.3 to 4.8).
Contrast between the experimental and the control interventions. We agree that the contrast between the experimental and the control interventions in our trial was relatively small. The main objective of our study was to investigate the specific effect of forced use, ie, the immobilization of the unaffected arm. We tried to make the control treatment equally intensive, to adjust for the nonspecific effects of treatment intensity. It is, indeed, conceivable that the intensity of the treatment is the most effective aspect of the treatment, designated as constraint-induced movement therapy, but this has not yet been shown in publications of well-designed studies in peer-reviewed journals. As we stated in the discussion paragraph of our article, the improvement in the control group may very well have been the result of the intensive physical and occupational therapy, which was equally intensive in the experimental group. The use of a splint and sling to immobilize the unaffected arm is very unpleasant and potentially dangerous. If a similar effect can be obtained by intensive bimanual treatment, this is of crucial importance for clinical practice. This would imply that the term “constraint-induced” can be replaced by “intensive.” The effectiveness of enhanced intensity of treatment for stroke patients has recently been shown by Kwakkel et al.R6
Validity and reliability of the MAL. The statement that the MAL is stable over a 2-week waiting period is not confirmed by the data presented by Miltner et al.R3 The MAL AOU and Quality Of Movement (QOM) data differ significantly between first contact and baseline (AOU, P=0.023; QOM, P=0.047), and between baseline and pretreatment (AOU, P=0.006; QOM, P=0.022), respectively (Wilcoxon signed rank test). In 2 of the articles claimed to confirm the stability of the MAL over a 6-week placebo treatment and follow-up period, no data were presented.R7 R8 More importantly, the possibility that reported changes on the MAL are the result of a Hawthorne effect remains unchallenged.
We hope that the readers, when weighing up the arguments in the letter from Taub and Uswatte and our reply to their criticism, will not forget that to date only 2 randomized controlled trialsR1 R2 on the question at issue have been published, of which oursR1 is by far the largest. We therefore wish to reemphasize the importance of our findings. We hope that the positive aspect of our conclusion, that the effect of forced use was clinically relevant in subgroups of patients with sensory disorders and hemineglect, will be a challenging starting point for future research.
van der Lee JH, Wagenaar RC, Lankhorst GJ, Vogelaar TW, Devillé WL, Bouter LM. Forced use of the upper extremity in chronic stroke patients: results from a single-blind randomized clinical trial. Stroke.. 1999;30:2369–2375.
Taub E, Miller NE, Novack TA, Cook EW III, Fleming WC, Nepomuceno CS, Connell JS, Crago JE. Technique to improve chronic motor deficit after stroke. Arch Phys Med Rehabil.. 1993;74:347–354.
Miltner WHR, Bauder H, Sommer M, Dettmers C, Taub E. Effects of constraint-induced movement therapy on patients with chronic motor deficits after stroke: a replication. Stroke.. 1999;30:586–592.
Kunkel A, Kopp B, Müller G, Villringer K, Villringer A, Taub E, Flor H. Constraint-induced movement therapy for motor recovery in chronic stroke patients. Arch Phys Med Rehabil.. 1999;80:624–628.
Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology: A Basic Science for Clinical Medicine. 2nd ed. Boston, Mass: Little, Brown & Co; 1995.
Taub E, Crago JE, Uswatte G. Constraint-induced movement therapy: a new approach to treatment in physical rehabilitation. Rehabil Psychol.. 1998;43:152–170.
Taub E, Wolf SL. Constraint-induced movement techniques to facilitate upper extremity use in stroke patients. Top Stroke Rehabil.. 1997;3:38–61.