Abstract
Background, Objectives, and Measurements: Patients with chronic airway obstruction (CAO) frequently experience dyspnea and fatigue during activities performed by accessory muscles of ventilation, which competitively participate in arm elevation. This systematic review of randomized controlled trials (RCTs) concerning patients with CAO addresses the effects of upper-extremity exercise training (UEET), added to lower-extremity training or comprehensive pulmonary rehabilitation, on the following patient-centered outcomes: exercise capacity, symptoms, ability to perform daily activities, and health-related quality of life.
Methods: Studies were retrieved using comprehensive database and hand-search strategies. Two independent reviewers determined study eligibility based on inclusion criteria. A detailed description of treatments was mandatory. Reviewers rated study quality and extracted information on study methods, design, intervention, and results.
Results: Forty publications were evaluated. Four RCTs met the inclusion criteria but had serious methodological limitations, which introduce possible biases that reduce their internal validity. The outcomes measured were heterogeneous, and the results were inconsistent regarding maximal exercise capacity, dyspnea, and health-related quality of life. No effect of UEET was demonstrated for measures of arm fatigue.
Limitations and Conclusions: The limited methodological quality of the studies retrieved prevented us from performing a meta-analysis, the results of which could be misleading. This systematic review shows that there is limited evidence examining UEET and that the evidence available is of poor quality. Therefore, a recommendation for the inclusion or exclusion of UEET in pulmonary rehabilitation programs for individuals with CAO is not possible. Further research is needed to definitively ascertain the effects of this training modality on patient-centered outcomes.
The upper extremities (UEs) play an important role in performing many activities of daily living (ADL), both in the domain of basic self-care and in everyday jobs. Patients with chronic airway obstruction (CAO) frequently experience marked dyspnea and fatigue when performing these tasks,1,2 which commonly require unsupported arm work and, therefore, pose a unique challenge to these individuals, whose upper-limb muscles are frequently recruited as accessory inspiratory muscles.3–5 During unsupported arm exercise, the participation of these muscles in ventilation decreases, and there is a shift of respiratory work to the diaphragm, which is commonly weakened and has a reduced functional capacity in these patients.6 This shift is associated with thoracic-abdominal desynchronization, severe dyspnea, and premature termination of exercise.1,7,8
The effectiveness of pulmonary rehabilitation (PR) programs has been well documented in patients with CAO, with consistent and clinically significant improvements in exercise capacity, symptoms, and health-related quality of life (HRQoL).9 However, such programs primarily focus on lower-extremity (LE) exercise training.10,11 Because training effects are specific to the limb trained, it seems reasonable to assume—but it so far remains unproven—that, in patients with CAO, upper-extremity exercise training (UEET) may improve functional status and reduce symptoms while performing ADL.
Recent guidelines from the American College of Chest Physicians11 recommend the introduction of unsupported endurance training of the UEs in PR programs. We think that this recommendation relies on limited evidence available from both randomized12–17 and nonrandomized18–20 studies conducted over recent decades. However, to our knowledge, no systematic review has ever been conducted on this topic.
We undertook this systematic review of randomized controlled trials (RCTs) to clarify the effect of UEET, implemented over and above standard treatment or lower-extremity exercise training (LEET), on patient-centered outcomes, such as exercise capacity, symptoms, ability to perform ADL, and HRQoL in patients with CAO.
Method
Data Sources and Searches
We performed a computer-based search, querying Ovid MEDLINE (1950 to March 2007), CINAHL (Cumulative Index to Nursing and Allied Health, 1982 to March 2007), EMBASE (1980 to March 2007), PEDro (Physiotherapy Evidence Database), and the Cochrane Central Register of Controlled Trials for original research articles published in English, Italian, and Spanish.
Search terms and strategies were as follows: (chronic airway obstruction OR pulmonary diseases chronic obstructive) AND (exercise therapy OR exercise OR rehabilitation OR physical therapy OR physiotherapy OR training) AND (arm OR upper extremities).
In addition, reference lists of relevant research articles were reviewed for pertinent studies. Abstracts presented at international meetings (American Thoracic Society, 2001–2007, European Respiratory Society, 2001–2007) also were hand-searched, and the authors of appropriate abstracts were contacted to obtain from them complete, unpublished data. Finally, PR experts were contacted to locate any further, unpublished material.
Study Selection
The following criteria were used to select trials for inclusion in the review.
Design.
We considered for inclusion RCTs only.
Target population.
Trials were considered when they enrolled patients with a diagnosis of moderate, severe, or very severe CAO. The criteria used for this purpose were the best recorded ratio of forced expiratory volume in 1 second (FEV1) to forced vital capacity (FVC) of less than 0.7, associated with the best recorded FEV1 of less than 80% of the predicted value,21 and a clinical diagnosis of CAO.
Intervention.
As recommended by the major scientific societies in this field,10 we selected any inpatient, outpatient, or home-based PR programs that included at least 20 sessions for a minimum frequency of 3 times a week. Both supervised and unsupervised home sessions were considered acceptable. The program had to include supported or unsupported UEET as the experimental intervention. A detailed description of the experimental intervention was mandatory.
Control.
Randomized controlled trials were included only when they compared UEET with treatments not specifically aimed at improving UE exercise capacity. The control group could receive standard training consisting of comprehensive inpatient, outpatient, or home-based PR programs, or it could be a training program targeting only LE exercise capacity. Again, the program had to include at least 20 regular sessions for a minimum frequency of 3 times a week; both supervised and unsupervised home sessions were acceptable. A detailed description of the control treatment was mandatory.
Outcome measures.
These measures could be arm exercise capacity (maximal exercise capacity, functional exercise capacity, or endurance time). The UE maximal exercise capacity was defined as the peak exercise capacity measured by an incremental exercise stress test. Functional exercise capacity was defined as the maximum number of UE elevations performed in 6 minutes, and arm endurance was defined as the duration of a constant-load, symptom-limited exercise, performed using an arm ergometer. Outcome measures also could be the symptoms of dyspnea or arm fatigue on exertion, which had to be quantified by specific, validated questionnaires as scores achieved during exercises requiring exerting the UEs; the ability to perform ADL tasks that involve the arms, using reliable measures; or the HRQoL, as assessed by data collected by specific questionnaires.
Data Extraction and Quality Assessment
To assess eligibility, 2 investigators (SC and EC) independently retrieved and examined the titles and abstracts of the studies in order to achieve higher accuracy in this process; a third investigator (RD) was consulted in case of disagreement to improve accuracy. The 2 investigators extracted the data from the studies selected for inclusion and requested important data missing from these reports from their authors. Finally, the 2 investigators independently rated the quality of the studies selected using the Consolidated Standards of Reporting Trials (CONSORT) statement.22
Results
Bibliographic Search Results
Forty publications were retrieved with combined computerized and hand search, including 2 abstracts and 1 reference (Figure). After a first review of the titles and abstracts, 13 studies remained potentially eligible. Efforts to obtain the full-length articles, or the data, directly from the authors of the reference and the 2 abstracts, presented at international meetings, were unsuccessful: in 2 cases, the authors did not answer our multiple attempts to contact them,20,23 and in the third case, the authors were unable to provide the information requested.24
Flow chart of screened, excluded, and eventually analyzed reports. Research articles included in more than one of the databases consulted (overlap) are not represented. RCT=randomized controlled trial, UEET=upper-extremity exercise training.
The 10 eligible full-text studies were independently reviewed as previously described. Agreement was reached to include 4 studies and to exclude 5 studies from the systematic review, using the Cohen coefficient of association (K=.8). One other study25 was excluded after consulting the third investigator. Summarizing, 6 research articles were excluded because they did not satisfy inclusion criteria for intervention or control. Two trials did not focus on UEET as the experimental intervention,25,26 and a third article was excluded because the intervention lasted less than 20 sessions,27 thus not satisfying the evidence-based criteria for duration of PR as stated by the American Thoracic Society/European Respiratory Society10 and because it included patients with diseases other than CAO. Two trials were excluded because the control groups did not perform comprehensive PR or LEET,14,15 and the sixth trial was excluded because it compared 2 different modes of UEET, thus not allowing for control.13
One trial selected for this review16 included 2 intervention groups, both satisfying our inclusion criteria, that were similar in the amount of training but different in the training modalities used. Both groups were separately considered for comparison with the control group. Another selected trial12 included 4 groups for multiple comparisons. We checked them for inclusion criteria and considered the group that performed LEET only for comparison with the group that performed UEET plus LEET.
Characteristics of the Samples
Table 1 shows baseline demographic characteristics of participants in the studies, as well as study design features of the 4 RCTs that fulfilled all of the eligibility criteria. The sample size was small in all of the trials (13–45 participants); altogether, 141 participants were randomized, and 107 participants completed the trials (76% of those who had been randomized). Among the patients who completed the trials, 60 were allocated to the intervention group and 47 were allocated to the control group.
Characteristics of the Patient Samples and Study Designs of Trials That Fulfilled Eligibility Criteriaa
Participants were elderly and had severe or very severe CAO.21 The main exclusion criteria in the selected trials were ischemic heart disease, heart failure, intermittent claudication, disabling musculoskeletal disorders, need for home oxygen treatment, hypercapnia, and medical conditions (other than CAO) severely limiting exercise tolerance.
Characteristics of the Training Programs
All of the selected trials included a detailed and complete description of the control and the intervention treatments (Tab. 2). Overall, UEET programs lasted 6 weeks16,17 to 8 weeks12,28 and included from 24 sessions12,28 to 55 sessions.16 In each study design, a short-term follow-up was planned, within 2 weeks from the end of the treatment.
Characteristics of the Experimental Intervention in the Trials Included in the Reviewa
Both the intervention and the control treatments were performed in an outpatient setting, although, in the study by Holland and co-workers,17 an unsupervised, daily home exercise program integrated the twice-weekly, outpatient, supervised sessions. Upper-extremity training was conducted as unsupported arm exercises against gravity and progressive resistance as the major component in all of the trials. A combination of unsupported and supported UEET, using an arm ergometer, was used in one trial.12
Outcome Measures
Table 3 summarizes the outcome measures used to assess the treatment effects. Arm exercise capacity (maximal exercise tolerance, functional exercise tolerance, or arm endurance time) was measured in all of the trials. An incremental stress test, with either supported UEs12,16 or unsupported UEs,17 was performed in 3 trials to assess the maximal exercise capacity, whereas 1 trial28 measured functional exercise capacity using a nonstandardized field test based on the number of arm elevations performed in 6 minutes. Finally, 1 trial16 measured the duration of a constant load exercise, sustained by patients on the arm cycle, at a work level one step below their previously determined maximum. We did not consider the submaximal test performed on an arm ergometer in the trial by Lake and co-workers12 because they did not report the endurance time.
Outcome Measures Used in the Trials Included in the Reviewa
Three trials16,17,28 measured dyspnea on exertion, and 3 trials12,16,17 measured arm fatigue on exertion, using different authorized versions of the Borg scale.29,30 One trial12 measured dyspnea with an unidentified scale; because the authors did not answer our request for clarification, the findings from this study are not reported for this specific outcome. The participants’ ability to perform several ADL tasks, predominantly involving the arms, was measured in 1 trial only,16 using a nonvalidated simulation field test. Three trials12,17,28 evaluated HRQoL, using self-administered questionnaires.31,32
Methodological Quality of the Included Trials
Quality of reporting of the studies selected was assessed by means of the CONSORT statement,22 which recently has been extended to cover reporting of nonpharmacological treatments such as physical therapy. It is a guideline designed to improve the reporting of RCTs. It consists of a checklist and a flow diagram that address the reporting of patient enrollment, allocation to treatments, follow-up, and data analysis. The CONSORT statement is widely used, and it is proven that the use of this evidence-based guideline is associated with improved quality of reporting of RCTs.33
The selected trials completely satisfied a minimum of 417 and a maximum of 712 of the 23 evaluation criteria included (Tab. 4). In all cases, an accurate explanation of the rationale and hypothesis of the study was given, as well as precise details of the treatments provided and the statistical methods used. Furthermore, the results were always interpreted in the context of current evidence. However, a number of criteria were only partially met or not at all satisfied, introducing the possibility of systematic errors that reduce the internal validity of trials.34 More specifically, in the trial of Ries and co-workers,16 eligibility criteria for patients were not specified, clear definitions of primary and secondary outcome measures were not provided, and there was not a clear description of the flow of participants through each stage of the trial. Furthermore, their results were weakened by the high number of patients who dropped out and whose data were not collected at follow-up.
Methodological Quality of the Trials Included in This Reviewa
Similarly, the trials by Lake et al12 and Sivori et al28 did not provide clear definitions of primary and secondary outcome measures, did not describe the flow of participants through each stage of the trial, did not provide the number of participants included in each analysis for each group, and did not specify whether the statistical analysis followed an intention-to-treat approach. Also in the trial by Sivori and co-workers,28 there was a large number of patients who dropped out and, therefore, were not reassessed at follow-up. Although Holland and co-workers17 provided a clear description of eligibility criteria for their patients, clarified the primary and secondary outcome measures, provided a blinded assessment of the outcome measured, described the flow of participants through each phase of the trial, and followed an intention-to-treat approach, they did not provide an estimate of the effect size with its precision (eg, 95% confidence interval) for any of the outcomes measured.
As a whole, none of the selected trials specified how sample size was determined or the methods used to generate and implement the random allocation sequence. In 3 trials,12,16,28 there was a complete lack of blinding for participants and the physical therapists administering the treatments and assessing the outcomes. One study17 added a placebo treatment to the PR performed by the control group to disguise their allocation to this group. The placebo treatment consisted of finger dexterity exercises performed in a sitting position with arm supported and, therefore, was not expected to improve arm exercise capacity. Furthermore, we would like to point out that 3 trials12,17,28 reported inequality in the male participant to female participant ratios, which might reduce the generalizability of their findings to the population of interest. The fourth trial16 did not report these data.
Because of the poor methodological quality of the 4 RCTs included, we decided not to perform a meta-analysis, the results of which could be misleading given the low internal validity of the trials. However, we based the conclusions of our review on the results of the between-group comparisons made in each of the selected studies, and we always considered their limits.
Effects of UEET
The effects of UEET, compared with conventional PR or LEET, are summarized in Tables 5 and 6. Maximal exercise capacity was measured in 3 trials, and data were obtained from 79 participants with severe or very severe CAO. In 2 trials,12,16 maximal exercise capacity was measured using an arm ergometer. In the third trial,17 maximal exercise capacity was quantified as the duration of a standardized field test,35 consisting of asking participants to raise their unsupported arms repeatedly, keeping an external pace, while the height of the target and the resistance were increased. Among the 3 trials mentioned, only 1 trial17 detected a statistically significant increment of the maximal exercise capacity in favor of the intervention (change score=55.3 seconds, 95% confidence interval [CI]=8.25 to 102.35, P<.02). One trial28 measured functional exercise capacity with a field test that satisfied our criteria. This outcome was collected in 28 individuals and documented a strong benefit in favor of the intervention group compared with the control group (change score=108, 95% CI=63.87 to 152.13, P<.0001), represented by an increased number of arm elevations per time unit. One trial16 measured endurance time by registering the duration of a constant-load, symptom-limited exercise performed using an arm ergometer, and it did not show any statistically significant difference between groups.
Effects of Upper-Extremity Exercise Training Plus Standard Training on Maximal and Functional Exercise Capacity, Endurance of the Upper Extremities, and Ability to Perform Activities of Daily Living (ADL) Involving the Upper Extremitiesa
Effects of Upper-Extremity Exercise Training Plus Standard Training on the Symptoms of Dyspnea and Arm Fatigue During Activities Involving the Upper Extremities and on Health-Related Quality of Life (HRQoL)a
Data regarding muscle effort of the UEs consistently showed no differences between intervention and control groups. Performance of ADL was measured in 28 patients with severe CAO enrolled in one trial.16 This measurement was collected with a nonstandardized field test, simulating 3 common ADL tasks that involve the UEs and are usually considered to be fatiguing in this population. No statistically significant difference was detected between groups in this domain, either in terms of the time required to perform activities or as perceived symptoms.
With regard to symptoms during exertion, 3 trials16,17,28 measured dyspnea and 3 trials12,16,17 measured the effort of the UE muscles. Altogether, data regarding dyspnea were collected in 94 patients, and 79 patients were assessed at follow-up for muscle effort. Although symptoms perceived during exertion always improved in both the intervention and control groups, a statistically significant difference in the dyspnea score, favoring the intervention group, was detected only in 1 trial28 (change score=−1.07, 95% CI=−1.87 to −0.27, P<.01). Two trials detected no benefits in favor of either group.
Health-related quality of life was measured in 3 trials12,17,28 for a total of 79 participants, using the Chronic Respiratory Disease Questionnaire31 in 2 studies and the Bandura scale32 in the third trial. Overall, the HRQoL improved for the intervention and control groups in all studies at the end of the treatment, but none of the studies revealed a statistically significant improvement in the intervention group compared with the control group.
Discussion and Conclusions
This systematic review demonstrates that there is insufficient evidence to support the inclusion of UEET in PR programs for patients with severe and very severe CAO. Although the results of 2 trials included,17,28 when considered separately, may suggest some advantages when UEET is incorporated into standard PR programs, the same results, when taken together, are strongly contradictory and, therefore, inadequate to recommend this activity.
Due to numerous shortcomings existing in the 4 RCTs included, the overall quality of the evidence collected in this systematic review was low for any of the outcomes studied. By examining the influence of key components of study quality for each trial reviewed, we found that potential sources of selection bias might exist in all of the selected trials. Sequence of allocation to treatments was not concealed in any of the RCTs, although it is well known that investigators’ knowledge of the sequence of allocation may cause selective enrollment of patients on the basis of prognostic factors34 and, consequently, may lead to inflated treatment effects.34
Performance bias and detection bias arise when the lack of double-blinding influences additional treatments that might be offered preferentially to one group or the assessment of the outcomes, respectively. Again, such biases may inflate treatment effects to a different degree, depending on the outcome assessed. This possibility is strongly reduced by the blinding of those administering the treatment, which is almost impossible in the field of physical therapy, and the blinding of patients and those assessing outcomes, which was accomplished by 1 trial17 among the 4 trials selected. The same trial was the only one that minimized the sources of attrition bias by making every effort to reduce the number of data lost to follow-up (1%) and using an intention-to-treat approach. Conversely, reporting of 3 trials12,16,28 was unclear regarding the approach followed in the data analysis, and in 2 trials16,28 the participant dropout rate was higher than 30%, thus reducing the validity of the findings. Moreover, not even one of the studies selected stated the intended sample size, and some trials needed multiple comparisons because they included more than one intervention group16,12 or control group,12 thus reducing the power of the analysis.
If UEET did indeed result in improvements, these changes may not have been able to be identified for several reasons, including the lack of sample power in all of these trials. Additionally, all of the trials implemented UEET specifically targeted at submaximal performance levels, whereas the testing procedures measured maximal performance. Finally, the benefit of UEET in the experimental group on the outcomes of HRQoL, dyspnea, and arm fatigue may have been masked by the concomitant participation in a PR program, which is known to improve these measures and would have done so in both experimental and control groups.
Thus, the overall quality of the trials included in this review is very low in 3 cases12,16,28 and unsatisfactory in the fourth case.17 Taking all of these weaknesses into consideration, the findings of this review cannot support the inclusion of UEET in PR programs for patients with CAO, and even the inconsistent advantages shown individually by some of the trials included12,17,28 may be overestimated.34
Notwithstanding these findings, the most recent guidelines for PR10,11 strongly recommend the introduction of unsupported UEET of sufficient duration (ie, 20 sessions) in PR programs. This was the main reason why we decided to include trials with at least 20 sessions of UEET in our review analysis, thus excluding other studies of different duration. Despite the belief that the longer the program, the greater the benefit,10 we cannot deny that significant benefits of UEET may occur in trials of shorter duration, such as in the study by Porta et al27; however, that particular study was performed in a very different population.
The rationale that supports the inclusion of specific training directed at the UEs in patients with CAO in PR programs relies on data from 6 randomized studies12–17 (3 included in this review) and 3 nonrandomized studies.18–20 The underlying principle is that an improvement in arm exercise capacity might be particularly important in these patients, whose UE muscles are competitively involved in both arm elevation and accessory ventilation. The same guidelines11 postulate that the mechanisms for improvement in UE function from such training include desensitization to dyspnea, better muscular coordination, and metabolic adaptation to exercise.
Numerous patients with stable, moderate to severe CAO complain of dyspnea during activities involving the UEs. These patients often show a characteristic association of dyspnea, dyssynchronous breathing,1 and inefficient metabolic and ventilatory response.1,36,37 A possible explanation of these phenomena is that, in patients with CAO who have hyperinflation, the diaphragm is less effective in performing inspiration.6 Consequently, during unsupported arm activities, these patients, compared with people who are healthy, must rely more on the accessory inspiratory muscles, which are involved in the competitive demands of ventilation and arm elevation. This fact poses greater demands on the accessory inspiratory muscle function, thus sustaining the hypothesis that these multifunctional muscles might benefit from specific training.38 However, the symptom of dyspnea was assessed in 3 trials16,17,28 based on unsupported UEET, and a statistically and clinically39 significant benefit in favor of the intervention was detected in only 1 trial.28 Taken together, the results reported in this review cannot support or refute the hypothesis that arm exercise may improve dyspnea.
The ultimate scope of rehabilitation is to improve the patient's autonomy in daily life. Pulmonary rehabilitation and exercise training, in particular, contribute in a decisive way to this process. Any accomplishment in this domain should be demonstrated by an increase in the patient's ability to perform ADL in his or her own environment. Unfortunately, the RCTs included in this review did not investigate these areas, and the trial16 that assessed the ability to perform ADL with a nonstandardized field test was unable to detect any favorable effects of UEET. Furthermore, arm fatigue was unchanged by the addition of UEET to a standard PR program. Interestingly, one trial screened and excluded from this review15 showed that, when UEET was administered independent of standard PR, it failed to provide any benefit compared with a control treatment. This finding suggests that UEET alone is not sufficient to improve clinically important outcomes for patients with CAO.
However, when implemented in the unsupported modality, UEET may add additional benefit to the established results of standard PR programs in terms of maximal and functional exercise capacity of the UEs. Indeed, Sivori and colleagues28 demonstrated a 100% improvement in functional exercise capacity of the arm, which was associated with a decrease in dyspnea but no change in HRQoL. Similarly, when a higher peak of exercise capacity was documented due to the effect of UEET, it did not translate to a reduction of dyspnea or arm fatigue, nor did it lead to a better HRQoL. Therefore, the effect of UEET on maximal exercise capacity is equivocal; findings in favor of UEET detected by individual trials are difficult to interpret and would exclude that desensitization to dyspnea and metabolic adaptation to exercise are possible mechanisms of improvement in UE exercise capacity, as recently postulated.11
This review did not demonstrate any additional improvement in HRQoL for patients who underwent UEET. In fact, among 3 trials12,17,28 that measured HRQoL, none showed a significant difference in favor of the intervention group. However, the trial by Lake and colleagues12 showed a trend favoring the intervention group in comparison with the control group (24% versus 7%). This benefit could be due to the Hawthorne effect, because in this trial patients were not blinded to treatment group allocation and no other improvement was detected in the other outcomes measured to substantiate this finding.
To our knowledge, this is the first systematic review examining the effectiveness of UEET in patients with CAO that has been performed using a rigorous, yet broad, search in different languages. The available evidence is limited, and the outcome measures examined varied considerably. Furthermore, the possibility that the samples were heterogeneous, coupled with the diverse UEET training protocols, limits the aggregation of the data. Finally, the relatively poor methodological quality of the included studies compromised both internal validity and generalizability of the results. These factors prevented us from conducting a meta-analysis, which might have been useful in clarifying the efficacy of UEET in patients with CAO.
In summary, the available evidence from RCTs appears inadequate to recommend in favor of or against the inclusion of UEET in PR programs for individuals with CAO. Further research should be conducted by means of well-designed and adequately powered trials, based on validated outcome measures addressed to clinically meaningful end points. The development of standardized and quantitative tests to assess the ability of patients with CAO to perform ADL also would be helpful to obtain a deeper understanding of clinically important achievements from the patients’ perspectives. Other important related research questions should be whether patients with different CAO severity or levels of disability might benefit differently from UEET and whether unsupported versus supported arm exercise might provide greater, or more selective, benefits.
Footnotes
-
Ms Costi, Dr Di Bari, Dr Crisafulli, Dr Fabbri, and Dr Clini provided concept/idea/research design. Ms Costi, Dr Di Bari, Mr Pillastrini, and Dr Clini provided writing. Ms Costi, Dr Di Bari, Dr Crisafulli, and Ms Arletti provided data collection. Ms Costi, Dr Di Bari, Dr D'Amico, and Ms Arletti provided data analysis. Ms Costi, Dr Di Bari, Mr Pillastrini, and Dr Crisafulli provided project management. Ms Costi, Dr Fabbri, and Dr Clini provided fund procurement, facilities/equipment, and institutional liaisons. Ms Costi, Dr D'Amico, and Dr Clini provided participants. The authors thank Mrs Jessie Cross for her helpful editorial advice and linguistic review of the manuscript.
- Received December 16, 2007.
- Accepted January 29, 2009.
- American Physical Therapy Association
References
Issue highlights
