Ottawa Panel Evidence-Based Clinical Practice Guidelines for Therapeutic Exercises in the Management of Rheumatoid Arthritis in Adults

Background and Purpose. The purpose of this project was to create guidelines for the use of therapeutic exercises and manual therapy in the management of adult patients ( (cid:1) 18 years of age) with a diagnosis of rheumatoid arthritis according to the 1987 American Rheumatism Association criteria. Methods. Evidence from comparative controlled trials was identified and synthesized using The Cochrane Collaboration methods. An expert panel was formed by inviting professional stakeholder organizations to each nominate a representative. This panel developed a set of criteria for grading the strength of both the evidence and the recommendation. Results. Six positive recommendations of clinical benefit were developed on therapeutic exercises. The efficacy of manual therapy interventions could not be determined for lack of evidence. Discussion and Conclusion. The panel recommends the use of therapeutic exercises for rheumatoid arthritis. Further research is needed to determine the efficacy of manual therapy in the management of this disease. [Ottawa Panel Evidence-Based Clinical Practice Guidelines for Therapeutic Exercises in the Management of Rheumatoid Arthritis in Adults. Phys Ther . 2004;84:934–972.]

Background and Purpose. The purpose of this project was to create guidelines for the use of therapeutic exercises and manual therapy in the management of adult patients (Ͼ18 years of age) with a diagnosis of rheumatoid arthritis according to the 1987 American Rheumatism Association criteria. Methods. Evidence from comparative controlled trials was identified and synthesized using The Cochrane Collaboration methods. An expert panel was formed by inviting professional stakeholder organizations to each nominate a representative. This panel developed a set of criteria for grading the strength of both the evidence and the recommendation. Results. Six positive recommendations of clinical benefit were developed on therapeutic exercises. The efficacy of manual therapy interventions could not be determined for lack of evidence. Discussion and Conclusion. The panel recommends the use of therapeutic exercises for rheumatoid arthritis. Further research is needed to determine the efficacy of manual therapy in the management of this disease. [Ottawa Panel Evidence-Based Clinical Practice Guidelines for Therapeutic Exercises in the Management of Rheumatoid Arthritis in Adults. Phys Ther. 2004;84:934 -972.] Introduction R heumatoid arthritis (RA) is a systemic inflammatory disease that produces a progressive degeneration of the musculoskeletal system. 1 One of the most prevalent chronic conditions, RA is found in approximately 1% of the adult population in the United States. [2][3][4][5][6] In adults, RA is more common among women than men by a ratio of 5:1 6 and is most prevalent among those aged 40 to 60 years. Rheumatoid arthritis is a highly disabling disease associated with high morbidity. Even with appropriate drug therapy, up to 7% of patients are disabled to some extent 5 years after disease onset and 50% are too disabled to work 10 years after onset. 7 Consequently, RA results in considerable direct costs, such as health care expenses, and indirect costs, such as loss of productivity due to morbidity and decreased life expectancy 1 ; these combined costs are estimated at 1% of the US gross national product. 8 Impairments, disabilities, and handicaps associated with RA can be devastating, leading to pain, activity restriction, and diminished quality of life, while placing a strain on the health care system and society. 1 Substantial progress has been made in the medical management of RA over the last decade, but rehabilitation specialists still must provide efficient and effective interventions for their patients. The development of evidence-based clinical practice guidelines (EBCPGs) for rehabilitation of adults with RA will help patients and clinicians choose effective interventions, which is important because the efficacy of rehabilitation interventions in RA management has a direct bearing on the combined costs of the disease. 6 According to Woolf, EBCPGs are "the official statements or policies of major organizations and agencies on the proper indications for performing a procedure or treatment or the proper management for specific clinical problems." 9(p1812) The appropriate use of such statements to direct practice has been proven beneficial to the rehabilitation process and patient health outcomes. 10 The Ottawa Panel was convened to evaluate the evidence for the effectiveness of 10 physical rehabilitation interventions for RA. Physical rehabilitation is a combination of therapeutic exercises, manual therapies, modalities, application of adaptive equipment, education, and re-education for the management of activities of daily living (ADL). The interventions examined by the Ottawa Panel were as follows: (1) acupuncture; (2) assistive devices; (3) bed rest; (4) conservation of energy; (5) electrotherapy, including electrical stimulation, lowlevel laser therapy, transcutaneous electrical nerve stimulation, and therapeutic ultrasound; (6) manual therapy; (7) patient education; (8) splinting and orthotics; (9) therapeutic exercises, with an emphasis on the intensity of the exercise program; and (10) thermotherapy, including heat therapy, cryotherapy, and balneotherapy. This article discusses only the evidence related to therapeutic exercises-including specific strengthening exercises and whole-body exercises (eg, general fitness and aerobic conditioning)-and manual therapy.
The target users of these EBCPGs for therapeutic exercises and manual therapy are physical therapists, occupational therapists, physiatrists, orthopedic surgeons, rheumatologists, family physicians, acupuncturists, and patients. The aim of developing the guidelines discussed in this article was to promote the appropriate use of therapeutic exercises and manual therapy in the management of RA.

Methods
The development process of these EBCPGs was similar to that of the Philadelphia Panel, except that a different target population was used. 11 Briefly, the Ottawa Methods Group (OMG), a group of 9 methodologists with experience in developing EBCPGs, asked professional associations interested in the care of people with RA for suggestions of individuals with both clinical expertise in the management of the disease and familiarity with EBCPGs. From among the suggestions given, the OMG chose 9 experts to serve as panel members. These experts in RA were a rheumatologist, a physiatrist, a physician with experience in evidence-based medicine, a family physician, 3 physical therapists (including one who practiced acupuncture and one involved in clinical research), an occupational therapist, and a patient with RA. The Ottawa Panel consisted of these 9 experts and all members of the OMG.
One OMG member assembled a research and support staff with expertise in meta-analyses, rheumatology rehabilitation interventions, research methods, or the development and assessment of EBCPGs. The OMG then established a priori a set of inclusion criteria for the study designs, subject samples, interventions, and outcomes to allow the research staff to select the most relevant material as evidence of the effectiveness of therapeutic exercise and manual therapy. The OMG also reviewed the inclusion criteria to ensure that the approach to the study selection was reproducible and systematic. This a priori protocol guided separate systematic reviews of the literature for each intervention.
The research staff reviewed articles and created evidence tables for them (see "Clinical Practice Guidelines"), which the 9 clinical experts received in preparation for their meeting with the OMG. These tables were used as the basis for making the recommendations.

Target Population
Included were studies with samples of adult patients (Ͼ18 years of age) with a diagnosis of RA according to the 1987 American Rheumatism Association (ARA) criteria. 12 A patient was said to have RA if he or she satisfied at least 4 of the following 7 ARA criteria: (1) morning stiffness, (2) arthritis of 3 or more joints, (3) arthritis of the hand joints, (4) symmetric arthritis, (5) rheumatoid nodules, (6) serum rheumatoid factor, or (7) radiologic changes. 12 Studies with patients with RA affecting peripheral joints were eligible. Studies with patients with both chronic and acute RA were included in our analysis because patients with both types of RA were included in the different clinical trials studied, sometimes in the same trial. Where possible, however, the recommendations clearly indicate whether the intervention is appropriate for chronic or acute conditions. The recommendations also include classification of functional capacity in patients with RA described as: (I) complete functional capacity with ability to carry out all usual duties without handicaps, (II) functional capacity adequate to conduct normal activities despite the handicap of discomfort or limited mobility of one or more joints, (III) functional capacity adequate to perform only a few or none of the duties of usual occupation or of self-care, or (IV) largely or wholly incapacitated, with the patient bedridden or wheelchair-bound, permitting little or no self-care. 6 When the recommendations do not indicate disease severity or functional severity, it is because the trial on which the recommendation was based did not mention severity (Appendix 1).
Studies of patients with RA who had back or neck problems were excluded because of the numerous and varied associated signs and symptoms. Another reason for not considering spine disorders for this article is that Philadelphia Panel guidelines developed by the same methodologists were recently published for back and neck pain. 11 Studies of patients who had recently had surgery also were excluded. Further exclusion criteria included studies with patients who had one of the following conditions: (1) other rheumatologic or musculoskeletal problems, such as tendinitis, bursitis, or fractures; (2) major medical problems that could interfere with the rehabilitation process or incapacitate functional status; or (3) psychiatric conditions. Studies of subjects without known pathology or impairments also were excluded. The majority of studies included patients with RA at chronic stages (Ͼ12 years' duration).
If the study sample contained individuals with mixed arthritic conditions, the study was excluded unless those conditions involved RA and osteoarthritis (OA), in which case the study was included only if the proportion of patients with RA was at least 75%. For further inclusion and exclusion criteria, see Table 1.

Literature Search
The library scientist developed a structured literature search based on the sensitive search strategy for randomized controlled trials (RCTs)-a strategy recommended by The Cochrane Collaboration 13 -and modifications proposed by Haynes et al 14 to that strategy. The Cochrane Collaboration method minimizes bias through a systematic approach to the literature search, study selection, and data extraction and synthesis. The search was organized around the condition and interventions rather than the outcomes because it was an a priori search. Thus, we had no control over the outcomes the authors decided to measure (see Appendix 2 for an example of the search strategy).
The library scientist expanded the search strategy to identify case-control, cohort, and nonrandomized studies and conducted the search in the electronic databases of MEDLINE, EMBASE, Current Contents, the Cumulative Index to Nursing and Allied Health (CINAHL), and the Cochrane Controlled Trials Register up to December 2002. She also searched the registries of the Cochrane Field of Rehabilitation and Related Therapies, the Cochrane Musculoskeletal Group, the Physiotherapy Evidence Database (PEDro), and the University of Ottawa EBCPGs Web site. Finally, she searched the reference lists of all of the included trials for relevant studies and contacted content experts for additional studies.
In the first round of study inclusion or exclusion, 2 independent reviewers, trained and experienced occupational therapist or physical therapist students, appraised the titles and abstracts of the literature search, using a checklist with the a priori-defined selection criteria (Tab. 1). More junior students were paired with fourth-year occupational therapist or physical therapist students who were experienced with the Philadelphia Panel 11 methodology. Each pair of reviewers was assigned to a specific intervention. Within each pair of reviewers, individuals independently read the title and abstract of each article and created an individual list of all of the articles of the database with a reason for including or excluding each article. If the reviewers were uncertain about a particular article after having read the abstract, they ordered the article and read it in full before making a determination. Before deciding whether to include or exclude the article, a comparison of their individual lists was performed. A senior reviewer who is a methodologist and a clinical expert in arthritis (LB) checked the 2 independent lists of articles and the reason for inclusion or exclusion to determine potential inconsistencies. Eleven percent of the abstracts reviewed needed the consultation of the senior reviewer. For the second round of inclusion and exclusion, the pairs of reviewers retrieved articles selected for inclusion from the first round and independently assessed the full articles for inclusion or exclusion in the study. Using predetermined extraction forms, the pairs of reviewers independently extracted data from included articles on the population characteristics, details of the interventions, trial design, allocation concealment, and outcomes. The pairs of reviewers assessed methodological quality using the Jadad scale, a 5-point scale with reported reliability and validity that assigns 2 points each for randomization and double blinding and 1 point for description of withdrawals. 15,16 The reviewers resolved differences in data extraction and quality assessment through consensus with the senior reviewer. This consensus served to support the reliability of data obtained with the article selection process.

Study Inclusion/Exclusion Criteria
The inclusion/exclusion criteria were based on previous criteria used by the Philadelphia Panel. This list of criteria, which had been created for multiple diagnoses, including back and neck pain, was adapted and approved by the OMG for use with RA (Tab. 1).
All original comparative controlled studies that evaluated the specific intervention in a sample of patients with RA were included: RCTs, controlled clinical trials (CCTs), cohort studies, and case-control studies. (Controlled clinical trials are the same as RCTs except that, according to the Jadad scale, CCTs are either not randomized or poorly randomized.) Crossover studies were included, and, to avoid potential confounders, the data from only the first part of the study (before crossing) were analyzed. (Data from the first part are more specific than data from the second part because once the study patients change from the intervention group to the placebo group, the outcome could be due to either the intervention or the placebo. Thus, such results are not useful for measuring the special effect of each intervention.) Uncontrolled cohort studies (studies with no comparison group) and case series were excluded, as were eligible studies with greater than 20% dropout rates or a sample size of less than 5 patients per group. Abstracts were excluded because none of the abstracts found had sufficient data for analysis and the full studies of the abstracts could not be obtained from the authors. Trials published in languages other than French and English were not analyzed because of the time and cost involved in translation. Head-to-head studies (that is, the comparison of 2 active interventions, such as therapeutic exercises versus transcutaneous electrical nerve stimulation) were generally excluded in these recommendations. Because we were interested in making a recommendation specifically about therapeutic exercise or manual therapy, we rejected head-to-head studies. At the meeting, the Ottawa Panel recommended that a direct com-parison of the intervention with either placebo or control was more valid for measuring the specific effect of the intervention. We did include, however, studies with head-to-head comparisons of high-versus low-intensity exercise as highly relevant for rheumatology practice in rehabilitation, especially in the presence of an inflammatory disease such as RA, where the dosage and intensity of therapy could make a difference in pain tolerance and joint damage. For further exclusion criteria, see Table 1.

Rehabilitation Interventions Related to Therapeutic Exercises and Manual Therapy
Rehabilitation interventions related to therapeutic exercises were identified as specific functional strengthening exercises, whole-body functional strengthening exercises, and physical activity. Strengthening exercises were defined as isometric, concentric, eccentric, and isokinetic resistance exercises. Specific functional strengthening exercises were defined as strengthening exercises applied to muscles crossing one specific joint or within one specific body part, such as the hand, shoulder, or knee. Whole-body functional strengthening exercises were defined as general strengthening exercises applied to muscles crossing many joints or within large body parts involving several joints such as the lower extremity. Physical activity was defined as a combination of strengthening and aerobic exercises (ie, therapeutic exercise and activities to increase endurance). Manual therapy was defined as passive physiologic and accessory joint movements, muscle stretching, and soft tissue mobilization applied to a specific joint. Definitions provided in this article were written according to the description of therapeutic exercises program in the primary trials included in this review (Appendix 1).
Acceptable comparators were placebo, untreated, or use of educational pamphlets or written instructions for self-management. Concurrent therapies (such as electroanalgesia and medication) were accepted only if provided to both the experimental and control groups. Studies with designs where patients were their own controls, were excluded. No limitations based on methodological quality were imposed a priori; however, the quality of the comparative controlled studies was considered when grading the recommendations resulting from our analysis.

Outcomes
The ўўўўўўўўўўўўўўўўўўўўўўўў based on the new proposal of the Canadian Society for the International Classification of Impairments, Disabilities, and Handicaps, 19 which involved the concepts of organic systems and impairment, abilities and disabilities, and life habits and handicap situation. The a priori outcomes were classified according to these concepts: (1) organic systems and impairment: number of inflamed joints, number of acute phase reactants (eg, erythrocyte sedimentation rate, which is "a test that measures the rate at which red blood cells settle through a column of liquid" 20 ), radiological damage, and side effects; (2) abilities and disabilities: pain reduction, muscle force, range of motion (ROM), postural status, and duration of morning stiffness; and ( The inclusion or exclusion of the report was determined by panel consensus. However, as many articles as possible were included to increase the statistical power of the final results. Each result comprised pooled data from studies measuring the same intervention and the same outcome over a similar time period.

Statistical Analysis
Data were analyzed using Review Manager software. 22 Continuous data, "data with a potentially infinite number of possible values along a continuum," 23 were analyzed using the weighted mean differences (WMDs) between the intervention and control groups at the end of the study, where the weight is the inverse of the variance. A WMD is "a method of meta-analysis used to combine measures on continuous scales (such as weight), where the mean, standard deviation, and sample size in each group are known." 23 Dichotomous data, or data with only 2 classifications, 23 were analyzed using relative risks. According to Cochrane, the relative risk is "the ratio of risk in the intervention group to the risk in the control group. The risk (proportion, probability, or rate) is the ratio of people with an event in a group to the total in the group." 23 Heterogeneity (ie, variability or difference between studies 23 ) was tested using the chi-square statistic. We tested data heterogeneity among the results of different included studies to make sure that only homogeneous data were pooled together. When heterogeneity was not significant, fixed-effect models were used. A fixed-effect model is a statistical model that stipulates that the units under analysis (eg, participants in a meta-analysis study) are the ones of interest and thus constitute the entire population of units. 23 Fixed-effect models were used to generalize data across the included studies. Randomeffects models include both within-study sampling error (variance) and between-studies variation in the assessment of the uncertainty (confidence interval) of a meta-analysis' results 23 and are more severe than fixedeffect models. Such random-effects models were used when heterogeneity was significant. All figures were created using Cochrane Collaboration methodology 22 (www.cochrane.org). The square in Figure 1 illustrates the WMD between the 2 groups when comparing them for a specific outcome of interest. The horizontal line represents the standard deviation of the WMD. If the standard deviation line touches the central vertical line of the graph, the confidence interval is 0 and the difference between the 2 groups is not statistically significant. For example, functional status, pain relief, or ROM in flexion for the group receiving shoulder strengthening exercises are not statistically different from those of the control group.
Based on previous studies in the musculoskeletal domain 24 and on consensus, 11 clinical improvement for all interventions studied by the Ottawa Panel was defined as 15% improvement relative to a control. This figure can be justified because it was developed by the Philadelphia Panel, whose members are experts in musculoskeletal practice, and confirmed by another panel (the Ottawa Panel) whose members included specialists in rheumatology and an expert biostatistician.
To determine clinical improvement, the absolute benefit and relative difference in the change from baseline were calculated. Absolute benefit was calculated as the improvement in the treatment group less the improvement in the control group, maintaining the original units of measurement. Relative difference was calculated as the absolute benefit divided by the baseline mean (weighted for the intervention and control groups). For dichotomous data, the relative percentage of improvement was calculated as the difference in the percentage of improvement between the intervention and control groups. 11 The recommendations were graded by their level (I for RCTs, II for nonrandomized studies) and strength (A, B, Cϩ, C, or D) of evidence. Evidence from one or more RCTs of a statistically significant, clinically important benefit (Ͼ15%) was necessary for a grade A recommendation. A grade B recommendation was given to a statistically significant, clinically important benefit (Ͼ15%) if the evidence was from observational studies or CCTs. Evidence of clinical importance (Ͼ15%) but not statistical significance earned a grade Cϩ recommendation. A grade C recommendation was given to those interventions where an appropriate outcome was measured in a study that met the inclusion criteria but no clinically important difference and no statistical significance were shown. Evidence from one or more RCTs of a statistically significant, benefit favoring the control group (Ͻ0%: favors controls) resulted in a grade D recommendation. Details on this grading system were published in the Philadelphia Panel methodology article. 11 Scales demonstrated to be valid and responsive to change are required to support a positive recommendation (A or B). Outcomes not supported in the scientific literature by an existing validation study but providing useful information in studies-such as morning stiffness duration and palm-to-pulp measurement of finger joint ROM-are insufficient to warrant a grade A or B recommendation. 17,18,25,26 Reviewing the Guidelines The guidelines were sent to the external experts for review. To judge clinical usefulness, the 20 positive recommendations also were sent to 5 practitioners for feedback. Practitioners were selected from clinical settings in the Ottawa and Toronto regions and were a physical therapist, an occupational therapist, a physiatrist, a family physician, and a rheumatologist, all of whom were currently working with patients with RA.
Practitioners were asked 4 questions for each guideline: whether the recommendation was clear, whether the practitioners agreed with the recommendation, whether they felt that the literature search on therapeutic exercises and intensity of rehabilitation was relevant and complete, and whether the results of the trials in the guidelines were interpreted according to the practitioners' understanding of the data. Results of this survey are shown in the "Results" section.

Literature Search
The literature search identified 2,280 potential articles on therapeutic exercises for several rheumatic conditions. Ninety of these articles were initially considered potentially relevant based on the selection criteria checklist for RA only. Sixteen of these articles relating to therapeutic exercises met the selection criteria and were included. 27-43 One of the 16 studies had a follow-up study, so we have counted these 2 studies as one (Tab. 2, Appendix 1). The other 74 trials 44 -117 were excluded from the final selection for various reasons (Tab. 3). For manual therapy, 862 articles were identified. Four of those articles were initially considered potentially relevant, but none were ultimately included 118 -121 (Tab. 4).

Therapeutic Exercises
The clinical practice guidelines for therapeutic exercises are shown in Appendix 3.

Summary of trials.
Sixteen trials (nϭ661 patients) evaluated different types of therapeutic exercises for RA affecting joints of the upper and lower extremities. All trials compared these exercises with a control, but the trials examined different kinds of exercise: (1) shoulder functional strengthening (nϭ28), 35 (2) hand functional strengthening (nϭ41), 32 (3) knee functional strengthening (nϭ35), 36 (4) whole-body functional strengthening (nϭ312), 28 -31,33,38 -41,43 (5) whole-body, low-intensity functional strengthening (group) that directly compared exercises with a home instruction program (nϭ100), 42 (6) physical activity compared with bed rest (nϭ145), 27,34,37 (7) whole-body, low-intensity (individualized) exercises versus written instructions received by a control group for a home exercise program (nϭ100), 42 (8) whole-body, high-intensity (group) exercises versus written instruction for a home exercise program (nϭ100), 42 and (9) whole-body, low-intensity (group) versus whole-body, high-intensity (group) exercises (nϭ100). 42 Six included trials were RCTs, 30,36,37,41-43 and 11 trials were CCTs 27-29,31-35,38 -40 (Appendix 1). We used the Jadad scale to decide whether a study was an RCT or a CCT. 11 In all trials, 2 main types of therapeutic exercises were prescribed: (1) muscle-specific functional strengthening exercises that included isometric, concentric, eccentric, and isokinetic resistance exercises and (2) whole-body functional strengthening programs that included general fitness and aerobic conditioning. The programs' durations ranged from 1 week to 6 months, the treatment schedule varied from 1 to 14 times a week, and the length of each exercise session ranged from 30 minutes to 1 hour (Appendix 1). Therapeutic exercises varied also in their extent of supervision (ie, supervised versus not supervised, group versus individual) and in their level of intensity (ie, low versus high).
Efficacy. Appendix 1 includes information on the intensity, frequency, and total duration of the exercises, which varied from study to study.
For shoulder functional strengthening versus control (one CCT, nϭ28), 35 no statistically significant difference or clinically important benefit was observed at 2 months for relieving pain or improving ADL and ROM in patients with chronic RA, functional class I or II, and shoulder pain (Fig. 1). No other outcomes were reported.
Hand functional strengthening versus control (one CCT, nϭ41) 32 showed no clinically important benefit for patients with chronic RA, functional class II or III, in improving ROM of the proximal interphalangeal (PIP) joint (results not shown) and grip force at 12 weeks (Fig. 2). However, hand functional strengthening did show a statistically significant difference (WMDϭϪ3.10°, 95% conference interval [CI]ϭϪ5.93°to Ϫ0.27°) with no clinically important benefit for PIP joint extension at 12 weeks only (Fig. 2).
A clinically important benefit (41% relative difference) was shown in knee functional strengthening versus control (one RCT, nϭ35) 36 for pain in patients who had seropositive or seronegative inflammatory RA and required long-term medication at 6 weeks (Tab. 5). No clinically important benefit was shown for function; no statistically significant difference was observed in any outcome measured after 6 weeks (Fig. 3).
For whole-body functional strengthening programs versus control (3 RCTs and 6 CCTs, nϭ312), 28 -31,33,38 -40,43 clinically important benefits were observed for swollen joints at 2 months (29% relative difference on the Lansbury's joint index), 39 number of sick leaves after 8 years (43%), 39 and quadriceps femoris muscle torque For low-intensity, whole-body functional exercises (individualized) versus a control group whose participants received instruction in a home-based program (one RCT, nϭ100), 42 statistically significant differences and clinically important benefits were obtained for change in function at 12 weeks (functionϭstatistically significant at 12  Whole-body, high-intensity exercises (group) versus control as described above (one RCT, nϭ100) 42 demonstrated no clinically important benefit for pain relief, muscle force, swollen/tender joints, joint mobility, or improvement in function (HAQ) at 3 and 6 months in patients with chronic RA (Tab. 12, Figs. 8a-b).
In the same RCT (nϭ100), 42 low-intensity supervised exercises (group) were compared with high-intensity exercises (group) and showed statistically significant differences and clinically important benefits for pain relief at 24 weeks (21% relative difference; WMDϭ 1.30 cm on a 10-cm VAS, 95% CIϭ0.20 -2.40 cm). Function only showed clinically important benefits at 12 weeks (HAQ; 21% relative difference; WMDϭ0, 95% CIϭϪ0.21 to 0.21). No clinically important effects were shown for muscle force, swollen/tender joints, or joint mobility at 3 and 6 months for patients with RA in a chronic stage (Tab. 13, Figs. 9a-b).

Strength of published evidence compared with other guidelines.
Good evidence (level I, RCT) exists that therapeutic exercises, including functional strengthening and low-or high-intensity exercises, relieve pain and improve overall function in patients with RA. The strength of evidence has been graded by the Ontario Program for Optimal Therapeutics, 122 which reported good-quality evidence related to therapeutic exercises (see Appendixes 4 and 5 for previous clinical practice guidelines on therapeutic exercises for RA and for shoulder pain). [123][124][125][126][127] Clinical recommendations compared with other guidelines. The Ottawa Panel concluded that good evidence exists (grade A for pain, function, and grip force; grade B for sick leave and lower-limb muscle force; grade Cϩ for swollen joints) that therapeutic exercises similar to those mentioned above, including functional strengthening and low-or high-intensity exercises, should be included as an intervention for patients with RA. Therapeutic exercises reduce pain while improving periarticular muscle force, aerobic capacity, and joint mobility (Appendix 4). This recommendation is in concordance with all other existing guidelines 122-125 and with 2 protocols. 128,129 Practitioners' response to Ottawa Panel guidelines. All practitioners surveyed agreed with the recommendations for therapeutic exercises. Two practitioners found the recommendations clear, while one practitioner was confused as to which intervention was effective. The Ottawa Panel responded that interventions with grades A, B, and Cϩ are effective depending on the specific outcome, and the summaries of the guidelines (see "Clinical Practice Guidelines") were rewritten to clarify this issue. The decision aid available on the University of Ottawa Web site (see below for more details) contributes to the clarity of the clinical application of the individual guideline.

Manual Therapy
Evidence with acceptable research design, interventions, group comparisons, or outcomes could not be identified to guide the development of recommendations for man-

Discussion
From this extensive systematic review, numerous EBCPGs (6 with grade A, B, and Cϩ recommendations) have been developed for therapeutic exercises for RA, including strengthening exercises and whole-body exercises, with an emphasis on intensity of the exercise program. One or more CCTs have shown that these interventions provide clinically important benefits. More evidence, however, is needed to determine the efficacy of therapeutic exercises and manual therapy in the management of RA (9 primary grade C recommendations for therapeutic exercises and one "insufficient data" for manual therapy). Although no harmful side effects were reported in the original studies, the results of our review seem to suggest some potentially negative effects of intervention. For example, physical activity as compared with bedrest may have negative effects on outcomes such as pain, function, ROM, number of tender or swollen joints, and time to walk 15.24 m (50 ft). The effects of high-intensity exercise on pain also raise concern.
However, as with all such reviews, this review has its limitations. The effectiveness of conservative management of patients with RA is a complex issue, 7 and rehabilitation specialists often use concomitant interventions in their daily practice. 130 For example, interventions such as cryotherapy, wax bath, and electrotherapy are used for pain relief or as treatment preparation before exercise intervention. The use of a single intervention does not reflect the complexity of the global approach adopted by rehabilitation specialists in real-life clinical situations.
Furthermore, the efficacy of therapeutic exercises for RA is thought to be influenced by a number of factors, 7 including biological, psychosocial, and environmental health indicators. 7,131 Therefore, a multidimensional clinical evaluation is recommended in arthritis management. 122,132 However, it was not possible to examine the effect of possible concurrent therapies such as medica-tion intake and thermotherapy on the effectiveness of the interventions examined. 133 The Ottawa Panel EBCPGs for the management of RA generally concur with previous and relatively recent EBCPGs for RA, 122-125 shown in Appendix 4, and with 2 protocols. 128,129 The Philadelphia Panel EBCPGs, on whose methodology those of the Ottawa Panel were based, were developed based on a systematic grading of the evidence determined by an expert panel. In both cases, the evidence was derived from new systematic reviews and meta-analyses conducted by the OMG using The Cochrane Collaboration methodology. The Ottawa Panel comprised several practitioners who verified the guidelines' applicability and ease of use for practicing clinicians. This additional procedure provides credibility for rehabilitation specialists who intend to use these EBCPGs in their daily practice.
The EBCPGs developed by the Ottawa Panel have some potential limitations due to methodological weaknesses. Although the included trials were selected based on well-established inclusion and exclusion criteria, selection was performed by occupational therapist and physical therapist students. Potential omission of studies due to reviewer inexperience could have led to selection bias. Consultation with a third reviewer (LB) and the use of the panel of senior clinical experts may have compensated in part for this potential methodological flaw. The EBCPGs also are limited by the inclusion and exclusion criteria for the included studies. For example, some reports of RCTs [31][32][33]36,37,40,41 did not specify if the study sample included individuals in acute or chronic stages of RA. Additionally, some studies lacked details about the specific characteristics of the exercise intervention such as intensity. This lack of specificity 18 could be problematic for future clinical implementation of the guidelines, especially for the whole-body functional strengthening recommendation.
The OMG, however, made sure that the development of the draft EBCPGs prepared for the expert members was in concordance with Appraisal of Guidelines Research and Evaluation (AGREE) criteria. 134 Using AGREE (www.agreecollaboration.org), 2 trained physical therapists assessed the Ottawa Panel EBCPGs for RA. This tool consists of 6 dimensions measured on a 4-point scale, where 1 represents "strongly agree" and 4 represents "strongly disagree." The dimensions are: (1) purpose, defined as overall objectives that described the potential impact of a guideline on society and populations of patients; (2) stakeholder involvement, defined as the extent to which the guideline represents the views of its targed users; (3) rigor of development, which deals with the process used to gather and synthesize the evidence and with the methods to formulate the recommenda- ўўўўўўўўўўўўўўўўўўўўўўўў tions and to update them; (4) clarity and presentation, which refers to the language and format of the guideline; (5) applicability, which relates to the likely organizational, behavioral, and cost implications of applying the guideline; and (6) editorial independence, which refers to the independence of the recommendations and acknowledgment of possible conflict of interest from the guideline development group. The EBCPGs obtained a very high score for dimensions 1 (purpose), 2 (stakeholder involvement), 4 (clarity), and 6 (editorial independence), with lower scores for dimensions 3 (rigor of development) and 5 (applicability). On the University of Ottawa School of Rehabilitation Sciences Web page (http://www.health.uottawa.ca/EBCpg/english/main.htm) precise results are currently available, and decision aids with detailed clinical application will soon be available. The rigor of development was low because of poor reporting of side effects and risks, which were not reported in the primary trials and therefore not included in the EBCPGs. The applicability was low, particularly in identifying potential organizational barriers, cost implications, and methods of applying and monitoring the guidelines. After publication, the Ottawa Panel is planning to implement these guidelines in the Arthritis Rehabilitation and Education Program of The Arthritis Society of Ontario.

Therapeutic Exercises
The Ottawa Panel concluded that therapeutic exercises, including specific functional strengthening and wholebody functional strengthening, are a beneficial intervention for patients with RA. The benefit may vary, however, according to disease acuity and the time frame during which the outcomes are measured. Clinical benefits are recognized for pain relief, upper-limb (grip) and lowerlimb force, and functional status. Other benefits include improved overall function and, of particular importance due to its socioeconomic impact, decreased number of sick leaves. In the presence of an inflammatory disease such as RA, a low-intensity exercise program favors the reduction of pain and an improved functional status as compared with a high-intensity program, which may exacerbate the inflammatory process and the risk of damage to the affected joints. This evidence was not reproduced in noninflammatory diseases such as OA. 135 Physiological changes in plasma opioid concentrations support the reduction of pain observed in patients with RA after exercise. 136 -138 The recommendation for therapeutic exercises is in concordance with all existing guidelines 122-125 and 2 protocols. 128,129 To our knowledge, all systematic reviews 112 and all existing descriptive literature 16,99,133,139 -142 support this recommendation. Some subtle variation exists, though, depending on the outcome studied.
Although the Ottawa Panel EBCPGs are based mainly on RCTs, further research investigating the efficacy of therapeutic exercises for patients with RA requires trials of higher methodological quality. Indeed, a large number of studies failed to meet the inclusion criteria. The overall methodological quality of the included studies underlying the EBCPGs was relatively weak 15 due to the difficulty in masking patients and evaluators for this kind of intervention. This methodolological weakness observed in the included RCTs may have caused an overestimation of effect. The impossibility of truly masking patients is a common problem in trials of rehabilitation interventions. 143 Additionally, although we found many RCTs on therapeutic exercises for RA, the authors did not always report the characteristics of the intervention, the characteristics of the sample, and the stage of the disease in a standardized way. Some outcomes studied in the primary trials may not be clinically plausible. For example, it is unclear how therapeutic exercise alone could lead to improvement in joint swelling. To improve methodological quality, future RCTs should use the Morin Theoretical Framework 18 and the CONSORT Model 144 to report not only the characteristics of clinical application, of the sample, and of the disease, but also of the dropouts, the method of randomization, and the use of validated measurements.
Investigators in future studies examining the benefits of therapeutic exercises in the management of patients with RA will need to be more explicit in specifying the characteristics of the implemented exercises and program, including aquatics programs 38,106,107 ; the intensity of the exercise; and the progression. In addition, to provide a more judicious evaluation of the benefits, patient-specific information concerning physical impair-

Manual Therapy
No studies of manual therapy with acceptable research designs were identified.

Implications for Practice
The Ottawa Panel found evidence to recommend and support the use of therapeutic exercises, especially knee functional strengthening, whole-body functional strengthening, general physical activity, and whole-body, low-intensity exercises, for the management of RA. Conversely, evidence is lacking at present as to whether the use of shoulder and hand strengthening exercises and whole-body, high-intensity exercises or manual therapy should be included or excluded in the daily practice of physical rehabilitation for RA management. It is important to note that the recommendations outlined here are limited by methodological considerations such as the quality of studies in the literature, including the generally poorly reported descriptions of therapeutic exercise programs, and the outcomes in those studies.

Figure 7b.
Low-intensity exercises (individualized) versus control (written instructions for home exercises).

Figure 8b.
High-intensity exercises versus control (written instructions for home exercises). HAQϭHealth Assessment Questionnaire.