To determine the effects of participation in a low-impact aerobic exercise program on fatigue, pain, and depression; to examine whether intervention groups compared with a control group differed on functional (grip strength and walk time) and disease activity (total joint count, erythrocyte sedimentation rate, and C-reactive protein) measures and aerobic fitness at the end of the intervention; and to test which factors predicted exercise participation.
A convenience sample of 220 adults with rheumatoid arthritis (RA), ages 40–70, was randomized to 1 of 3 groups: class exercise, home exercise using a videotape, and control group. Measures were obtained at baseline (T1), after 6 weeks of exercise (T2), and after 12 weeks of exercise (T3).
Using structural equation modeling, overall symptoms (latent variable for pain, fatigue, and depression) decreased significantly at T3 (P < 0.04) for the class exercise group compared with the control group. There were significant interaction effects of time and group for the functional measures of walk time and grip strength: the treatment groups improved more than the control group (P ≤ 0.005). There were no significant increases in measures of disease activity. Fatigue and perceptions of benefits and barriers to exercise affected participants' amount of exercise, supporting previous research.
This study supported the positive effects of exercise on walk time and grip strength, and demonstrated that fatigue and perceived benefits/barriers to exercise influenced exercise participation. Furthermore, overall symptoms of fatigue, pain, and depression were positively influenced in this selective group of patients with RA ages 40–70 years.
Arthritis and other rheumatic diseases are leading causes of disability in the US and are predicted to affect 60 million people by 2020 (1). Rheumatoid arthritis (RA) may hasten functional decline over and above that associated with aging. Fatigue, pain, and depression, hallmark symptoms of RA (2), may increase poor quality of life and functional decline. Dynamic forms of exercise increase muscle strength and endurance and help prevent aging-related functional decline (3). Most persons with RA have quite poor levels of aerobic fitness (4). In a prospective study of older women (n = 6,632), slow gait, weak grip, and limited exercise, all common in RA (3), were among modifiable predictors of functional decline (5).
The Cochrane Review (6) assessed 6 randomized clinical trials (RCTs) of dynamic exercise in persons with RA (7–12): only 2 RCTs included fatigue (9, 12) and only 2 included depression (11, 12) as outcome measures. None described treatment allocation precisely, the median sample size was 32, only 2 performed an intent-to-treat analysis (7, 12), 2 had blinded assessors (8, 10), and 2 did not include full weight-bearing exercises (7, 9). The Cochrane Review concluded that dynamic exercise lasting 3 months or less is effective in increasing aerobic capacity, muscle strength, and joint mobility, with no detrimental effects on disease activity or pain, and that effects of dynamic exercise on function and radiologic progression are unclear. Assessment of additional health status measures, including fatigue, in future exercise studies was recommended (6). Another review of RCTs of aerobic and strengthening exercises also supported positive effects of exercise on RA (13).
Little is known, however, about factors that influence exercising in persons with RA. Some studies found that perceived benefits of exercise and self-efficacy significantly predicted exercise among patients with RA (4, 14). Other studies found that history of exercise and the rheumatologist's current exercise behavior (15) were exercise predictors. Social support for exercise and prior exercise behavior (16) and postexercise perceptions of feeling better (17) also have been found to be important for exercise adherence.
We based our study on self-regulation theory, which posits that individuals use goals to guide their behavior. If a discrepancy is noted between a goal and one's present state, behavior is adjusted. If difficulty is encountered in adjusting behavior, one assesses whether further effort will achieve the goal. Outcome expectancy is the term used to describe this assessment (18). According to self-regulation theory, psychosocial variables (social support, self-efficacy for exercise, barriers to and benefits of exercise, and optimism) influence outcome expectancy assessments (19).
One large (n = 309) RCT of high-intensity activities (70–95% of maximum heart rate [MHR]) versus usual care demonstrated greater improvement in functional ability in exercise participants (20). In contrast to that study of high-intensity exercise, our study tested predictors and effects of low-impact aerobic exercise, which is more realistic for individuals with RA (21). We included a third group who exercised at home using a videotape, which if effective may more likely result in continued exercise. No other RCTs were found that tested outcomes of class exercise compared with home exercise and a no-exercise control group. Nor were any studies found that tested a theoretical model of predictors of exercise during an exercise intervention.
Based on previous literature and a one-group pilot study that found increased aerobic capacity and grip strength and decreased pain, fatigue, and walk time after 12 weeks of exercise (22), we hypothesized that compared with a no-intervention control group, 1) participation in a class low-impact aerobic exercise program (C-Tx) or a home exercise program using a videotape (H-Tx) would result in less fatigue, pain, and depression among outpatients with RA, and 2) participation in a C-Tx program or an H-Tx program using a videotape would improve function (grip strength and walk time) and aerobic fitness, with no differences in disease activity (erythrocyte sedimentation rate, C-reactive protein [CRP] level, and total joint count). In addition, we explored the following research question: What baseline (T1) psychosocial (optimism, exercise benefits/barriers, social support, self-efficacy for exercise), functional, aerobic fitness, symptoms, and disease activity factors predict exercise participation among the 2 intervention groups?
After institutional review board approval, participants were recruited through an arthritis clinic, private rheumatologists, media advertisements, and posted flyers. Between 1996 and 1999, potential participants (n = 789) were assessed for eligibility (Table 1) and 404 were invited to participate (Figure 1). The 310 who met the inclusion criteria, gave verbal consent, and had RA diagnosis confirmed (23) were randomly assigned to 1 of the 3 groups, stratified by sex using an a priori list of randomly generated permutations of 3 numbers: 1 = C-Tx, 2 = H-Tx, and 3 = control. A total of 220 participants completed the study (68 in the C-Tx, 79 in the H-Tx, and 73 in the control group). An a priori power analysis (24) identified 53 participants per group as sufficient to determine treatment effects using analysis of variance (ANOVA; α = 0.05, medium effect 0.25, power of 0.80). This was increased for attrition and SEM analysis, resulting in a need for complete data on 216 participants.
|
| Age 40–70 years |
| Confirmed RA diagnosis according to ACR (formerly the ARA) 1987 criteria (23) |
| Able to read and speak English |
| Ambulatory |
| No history of fibromyalgia or severe COPD |
| Not taking a beta-blocker or digitalis medication |
| Not presently performing ≥30 minutes of aerobic exercise ≥3 times weekly |
| Have rheumatologist/physician approval to participate |
| Able to meet criteria for aerobic fitness testing (no arrhythmias, recent myocardial infarction, acute infection, uncontrolled metabolic disease, known electrolyte abnormalities, or systolic BP >200 mm Hg or diastolic BP >115 mm Hg) |
Flow chart of randomized clinical trial. T1 = baseline assessment; T2 = 6-week assessment; T3 = 12-week assessment; RA = rheumatoid arthritis; COPD = chronic obstructive pulmonary disease.
Sixty-two participants dropped out or were determined as no longer eligible before attending the first assessment session (T1), which was scheduled several months after initial recruitment, and 28 dropped out after T1 (Figure 1). Although we had no data on the 62 participants who dropped out before T1, we compared the 28 who dropped after T1 with those who completed the study on all variables. The only differences were that dropouts after T1 were more likely to be minorities (21% versus 9%), had RA for more years (16.4 years versus 11.2 years), and took longer to walk 50 feet at T1 (11.6 seconds versus 9.7 seconds). A check on random assignment confirmed no significant group differences at T1 for functional, disease activity, or aerobic fitness measures.
Written consent was obtained at T1. Measures were obtained in clinic rooms at baseline (T1), midtreatment (T2), and after the 12-week intervention (T3). All trained assessors were blinded to participants' group assignment.
The intervention was 12 weeks of low-impact aerobic exercises for 1 hour 3 times a week. In low-impact aerobic exercise one foot is always on the ground and there are no running or jumping movements. C-Tx participants attended classes at a fitness center; H-Tx participants exercised at home using a videotape of the same exercise program. The exercises consisted of 4 phases: warm-up, low-impact aerobics, strengthening, and cool-down exercises. Distribution of minutes for each phase of exercise (warm-up, aerobics, strengthening, and cool-down, respectively) was as follows: 20, 10, 20, and 10 for week 1; 15, 20, 15, and 10 for weeks 2–3; 10, 25, 20, and 5 for weeks 4–6; and 10, 30, 15, and 5 for weeks 7–12.
Each treatment group participant was given their target heart rate for 60% and 80% of their MHR and instructed to start exercising at 60% and progress to 80% as tolerated, given their subjective exertion using the Talk Test (being able to talk while exercising without being short of breath) and the Borg scale (25). The instructor encouraged exercising at targeted levels, both in class and on videotape. Participants were trained to take their pulse, prompted when to do so, and recorded pulse rates after the aerobic portion of each session. All participants reported the type and amount of other exercise performed during the past week at each assessment. The control group participated in all assessments and was asked to keep exercise levels at baseline amounts.
Cronbach's alpha, a measure of internal consistency reliability, was computed with T1 data for all scales except the 1-item pain measures, and all were ≥0.81 (range 0.81–0.96). All measures except social support, the Life Orientation Test, and self-efficacy were used in our pilot study (22).
Information on age, sex, race, education, income, marital status, work status, RA duration, and comorbidities were obtained at T1.
Fatigue was measured with the 14-item Global Fatigue Index of the Multidimensional Assessment of Fatigue questionnaire; a higher score indicated greater fatigue (26). Measures of validity have been reported (26, 27).
Pain was measured with the Short Form of the McGill Pain Questionnaire (28). Good reliability and validity have been reported (29). In this study, the 1-item Present Pain Intensity index and a 1-item pain description scale were used.
Depression was measured with the Center for Epidemiologic Studies Depression Scale. Responses to 20 statements were summed for a total score, with >16 indicating depressive symptoms (30). The 30-item Profile of Moods States Short Form, a valid and reliable measure of mood disturbance (31), was used for additional measures of depression and fatigue.
Disease activity was measured using total joint count (TJC), erythrocyte sedimentation rate, and CRP level (normal value = 0 mg/dl). TJC is the number of tender and swollen joints (32) and is higher when RA is active. The same trained research assistant (a clinical nurse specialist) assessed each participant at each assessment period by palpating each joint. Erythrocyte sedimentation rate and CRP level measured RA inflammation (32) using standard laboratory procedures (33).
Function was measured with grip strength and walk time assessed by a physical therapist (PT) or trained PT student. Grip strength, a recognized functional measure in RA, has high test–retest reliability and criterion validity (34). A blood pressure cuff connected to a portable sphygmomanometer was pumped to 20 mm Hg. Participants squeezed the cuff as hard as possible 3 times with each hand. A mean of the 3 measures was used in the analyses.
Walk time, the number of seconds it takes to walk 50 feet, measured with a stopwatch, is another functional measure in RA (35) with established reliability and validity (36). Each year a PT and trained PT students simultaneously assessed walk tests of 5 randomly selected participants, resulting in high interrater reliability.
Bicycle ergometer testing, using the Astrand-Rhyming protocol, provided data for calculating estimated maximum oxygen consumption (VO2max) for each participant (25). Estimated VO2max and maximal testing are highly correlated (r = 0.94) (37). The protocol, conducted at T1 and T3 by a PT, involved pedaling on the cycle for 6 minutes or until a steady-state heart rate was achieved. A cardiac monitor with 5 electrodes monitored heart rates. Either an all-extremity cycle or an arm cycle test was used for 12 participants who had difficulty cycling. The all-extremity cycle also correlates highly (r = 0.91) with maximal testing (38) whereas the arm cycle is 25% lower than maximal testing (39).
Expectations about positive outcomes in life were measured with the Life Orientation Test at T1 and T3 (40). A higher score indicates greater optimism and acceptable internal consistency (40) and test–retest reliability have been reported (41).
The 43-item Exercise Benefits/Barriers Scale (42) measured perceived benefits (29 items) and barriers (14 items) to exercising. With reverse scoring barrier items, a higher total score indicates more perceived benefits and fewer perceived barriers.
Social support was measured with the Medical Outcomes Social Support Survey, an 18-item questionnaire that measures functional social support and has demonstrated validity (43). Self-efficacy was measured with the mean of 2 confidence measures, a 9-item exercise self-efficacy scale with reported concurrent validity (44) and a 5-item measure of confidence in one's ability to persist with exercising when potential barriers exist (45). Construct validity has also been reported (45). Higher scores indicate greater self-efficacy.
For the treatment groups, amount of exercise was calculated by summing mean minutes per week of the aerobic portion of the intervention (half of each 1-hour session completed) and the mean minutes per week of other aerobic exercise reported at T2 and T3. Control group participants also reported the number of minutes of aerobic exercise per week at T2 and T3.
Participants reported all medications at T1 and any medication changes at subsequent assessments. Arthritis medications were coded as either nonsteroidal antiinflammatory drugs or disease-modifying antirheumatic drugs (DMARDs).
Intent-to-treat was used in all analyses. Descriptive statistics were obtained for all variables using SPSS software, version 8.0 (SPSS, Chicago, IL). Chi-square or ANOVA was used to test for differences in baseline variables among the groups. Missing data were limited and random. In all cases, at least 80% of the items were needed to calculate scale scores. Occasionally, participants' mean scores on a daily measure (completed 1 week following each assessment but not otherwise used in this study) were used to replace a missing value. Means, standard deviations, medians, and ranges by group at all assessments are shown in Tables 2 and 3.
| T1 | T2 | T3 | |||||||
|---|---|---|---|---|---|---|---|---|---|
| C-Tx | H-Tx | Control | C-Tx | H-Tx | Control | C-Tx | H-Tx | Control | |
| |||||||||
| Global fatigue (MAF) (possible range 3–50) | 24.91 ± 10.25 (7.5–47.91) | 20.08 ± 10.15 (4.0–46.45) | 21.88 ± 9.80 (4.0–44.18) | 23.38 ± 11.57 (4.00–46.73) | 19.52 ± 9.11 (4.00–44.64) | 20.04 ± 11.34 (4.00–49.78) | 20.74 ± 11.61 (4.0–47.09) | 19.23 ± 10.55 (4.0–48.30) | 20.88 ± 11.20 (4.0–47.22) |
| POMS fatigue (possible range 0–4) | 1.53 ± 0.98 (0–3.80) | 1.16 ± 0.81 (0–3.20) | 1.52 ± 0.82 (0–3.40) | 1.61 ± 1.05 (0–4.00) | 1.29 ± 0.81 (0–3.20) | 1.42 ± 0.92 (0–4.00) | 1.35 ± 1.08 (0–4.00) | 1.23 ± 0.91 (0–3.60) | 1.37 ± 1.03 (0–4.00) |
| McGill Pain Intensity (possible range 1–10) | 4.67 ± 2.14 (1–9) | 3.88 ± 1.90 (1–10) | 4.14 ± 2.31 (1–9) | 4.98 ± 2.34 (1–10) | 4.56 ± 2.12 (1–9) | 4.49 ± 2.26 (1–10) | 4.05 ± 2.24 (1–10) | 4.15 ± 1.94 (1–9) | 4.34 ± 2.25 (1–10) |
| McGill description of pain today (possible range 0–5) | 2.00 ± 1.02 (0–5) | 1.71 ± 0.84 (0–4) | 2.01 ± 1.11 (0–5) | 2.16 ± 1.14 (0–5) | 1.87 ± 0.90 (0–4) | 2.07 ± 0.95 (0–4) | 1.67 ± 1.13 (0–5) | 1.78 ± 0.88 (0–4) | 1.95 ± 1.03 (0–5) |
| CES-D depression (possible range 0–60) | 14.81 ± 8.12 (0–35.79) | 10.62 ± 7.74 (0–35) | 12.86 ± 8.57 (0–41) | 13.74 ± 8.17 (2–46) | 10.49 ± 7.67 (0–36) | 12.79 ± 8.30 (0–35) | 13.74 ± 9.46 (0–37) | 10.45 ± 8.16 (0–40) | 11.65 ± 9.00 (0–37) |
| POMS depression (possible range 0–4) | 0.59 ± 0.67 (0–2.80) | 0.46 ± 0.59 (0–2.50) | 0.62 ± 0.63 (0–2.60) | 0.47 ± 0.58 (0–2.60) | 0.44 ± 0.60 (0–3.00) | 0.42 ± 0.53 (0–2.60) | 0.49 ± 0.62 (0–3.20) | 0.36 ± 0.67 (0–3.50) | 0.42 ± 0.62 (0–2.40) |
| Total joint count (possible range 0–164) | 32.16 ± 29.07 (0–117) | 29.04 ± 23.09 (2–104) | 37.14 ± 25.13 (0–112) | 32.81 ± 31.68 (0–122) | 23.19 ± 21.90 (0–98) | 32.99 ± 25.76 (0–122) | 31.04 ± 29.21 (0–114) | 23.71 ± 23.07 (0–107) | 35.14 ± 26.72 (0–115) |
| ESR (mm/hour) | 32.47 ± 22.71 (1–101) | 23.68 ± 25.57 (1–121) | 27.59 ± 24.09 (1–120) | 30.78 ± 22.10 (2–102) | 24.19 ± 24.73 (1–123) | 29.68 ± 25.59 (1–100) | 31.99 ± 24.17 (1–105) | 21.90 ± 21.85 (0–113) | 26.79 ± 23.43 (1–110) |
| CRP (mg/dl) | 1.33 ± 2.00 (0.4–11.1) | 1.25 ± 1.58 (0.4–8.2) | 1.30 ± 1.76 (0.4–9.6) | 1.20 ± 1.47 (0.4–6.8) | 1.07 ± 1.25 (0.4–8.4) | 1.14 ± 1.42 (0.4–9.5) | 1.13 ± 1.36 (0.4–6.1) | 0.86 ± 1.03 (0.4–7.0) | 1.02 ± 1.21 (0.4–6.0) |
| Left grip strength (mm hg; possible range 20–260) | 117.40 ± 46.82 (20–236.67) | 134.65 ± 59.42 (41.33–260) | 134.84 ± 56.56 (29.33–260) | 131.44 ± 52.71 (26.00–256) | 142.23 ± 61.09 (42.67–260) | 139.63 ± 60.74 (28.67–260) | 138.82 ± 54.58 (20–260) | 144.67 ± 63.78 (36.67–260) | 138.07 ± 59.51 (20–260) |
| Right grip strength (mm hg; possible range 20–260) | 121.37 ± 52.35 (24.67–260) | 130.94 ± 58.72 (32.67–260) | 133.38 ± 58.65 (30.67–260) | 134.89 ± 57.09 (38.00–260) | 141.39 ± 61.54 (39.33–260) | 141.09 ± 62.04 (25.33–260) | 141.84 ± 56.63 (44.67–260) | 144.81 ± 64.86 (31.33–260) | 142.95 ± 60.31 (34.00–260) |
| Walk time (minutes) | 10.04 ± 3.11 (5.44–22.02) | 9.64 ± 5.19 (5.21–50.47) | 9.36 ± 2.83 (4.95–23.89) | 9.57 ± 2.60 (4.89–17.46) | 9.09 ± 2.93 (3.31–24.41) | 10.08 ± 4.68 (5.41–42.56) | 9.33 ± 2.81 (3.43–17.88) | 9.40 ± 4.37 (4.21–38.77) | 9.97 ± 3.88 (5.13–32.13) |
| T1 | T2 | T3 | |||||||
|---|---|---|---|---|---|---|---|---|---|
| C-Tx | H-Tx | Control | C-Tx | H-Tx | Control | C-Tx | H-Tx | Control | |
| |||||||||
| Optimism (possible range 0–4) | 2.71 ± 0.67 (1.25–3.88) | 2.76 ± 0.61 (1.50–4.00) | 2.66 ± 0.61 (1.25–4.00) | NA | NA | NA | 2.76 ± 0.61 (1.50–4.00) | 2.86 ± 0.57 (1.13–4.00) | 2.67 ± 0.60 (1.38–4.00) |
| Exercise benefits and barriers (possible range 43–172) | 130.43 ± 13.35 (104.00–167.90) | 130.95 ± 15.81 (102.00–166.00) | 127.25 ± 14.49 (89.00–167.00) | 136.27 ± 14.72 (100.68–172.00) | 134.98 ± 16.56 (102.38–171.00) | 127.41 ± 17.14 (86.00–169.00) | 135.75 ± 13.51 (100.33–170.00) | 133.02 ± 16.34 (94.19–169.00) | 127.32 ± 14.62 (87.00–172.00) |
| Self-efficacy (possible range 0–7) | 5.04 ± 1.33 (2.08–7.00) | 5.38 ± 0.98 (2.46–7.00) | 4.87 ± 1.16 (2.54–7.00) | 5.35 ± 1.10 (1.62–7.00) | 5.30 ± 1.01 (2.62–7.00) | 4.86 ± 1.26 (2.15–7.00) | 5.31 ± 1.05 (2.38–7.00) | 5.13 ± 0.99 (2.85–7.00) | 4.80 ± 1.31 (0.92–7.00) |
| Amount of exercise (minutes/week aerobic exercise) | NC | NC | NC | NC | NC | NC | 85.45 ± 36.30 (5.00–240.00) | 92.54 ± 45.37 (17.50–296.67) | 29.81 ± 33.22 (0–131.67) |
| Aerobic fitness (VO2max in ml/kg/minute) | 22.50 ± 9.15 (9.17–54.40) | 23.32 ± 7.19 (4.91–39.90) | 21.10 ± 8.15 (5.28–42.50) | NA | NA | NA | 25.09 ± 9.25 (9.36–49.06) | 24.58 ± 8.07 (5.64–45.07) | 22.71 ± 7.85 (4.44–43.63) |
Structural equation modeling (SEM), a specialized parametric regression technique (46), was used to answer the first hypothesis. Quantile-quantile plots confirmed normal distribution of data. Each of the 3 symptoms was assessed with 2 measures so that a latent variable for each symptom was estimated. The overall fit of the SEM model was assessed using the Confirmation of Fit Index (CFI) (46). A CFI score <0.90 is considered inadequate, 0.90–0.94 is considered adequate, and >0.95 is considered very good.
The advantage of SEM is that it accommodates multiple measures of concepts and generates latent or unmeasured variables. For example, instead of determining the effects of 2 measures of depression, or deciding between them or how to combine them if they are highly correlated (as would be likely), the 2 measures serve as indicators of a latent variable for depression in the model. This is a stronger measure of the concept of interest because it takes advantage of more than 1 empirical measure and adjusts for measurement error when calculating relationships involving the latent variable. Finally, latent variables for overall symptoms were estimated combining fatigue, pain, and depression. The overall symptoms score is essentially a weighted average of the individual symptom scores. The weights and calculation of overall symptoms are automatically determined in SEM. Group assignment was dummy coded.
To answer the second hypothesis, after determining normality of data, repeated-measures ANOVA was conducted to determine if there were significant group differences in the functional, disease activity, and aerobic fitness measures over time, or for the interaction of time and group. Repeated-measures ANOVA was used because we did not have multiple measures of these variables in order to take advantage of and use SEM as in the first hypothesis.
A classification and regression tree (CART) analysis was used to answer the research question about predictors of exercise. CART is a nonparametric regression model that makes no assumptions about distribution of errors (47). It provides a flexible exploratory analysis and allows identification of interactions among variables. CART does not provide typical inferential measures such as P values. Fitting a CART model results in a large tree with sets of nodes. Each node represents a partition of the dependent variable from binary cut points in the independent variables. Iterative process reduces the number of nodes, resulting in a more parsimonious model. In the CART analysis, T1 measures of symptoms (fatigue, pain, depression), psychosocial measures (social support, self-efficacy for exercise, perceived benefits of exercise, and optimism), demographics (age, education, sex, duration of RA, number of comorbidities), functional measures (grip strength, walk time), aerobic fitness, and disease activity measures (erythrocyte sedimentation rate, CRP level, total joint count) were examined as predictors of minutes of exercise.
The median age of the 220 participants was 55.5 years (range 40–70 years) and the median years with RA was 8.0 (range 0.5–50). Participants were predominantly white (85%), 9.5% were African American, 4.1% were Hispanic, and 1.4% were Native American. The median education category was some college, 82.7% were women, most (70%) were married, 40% worked full time and 15.9% worked part time, and the median income level was $40,000–$49,999. The only significant group differences in demographics were for race (χ2 = 22.16, P < 0.001; more minorities in the C-Tx group than in the H-Tx group) and educational level. The H-Tx group had higher education than the control group (χ2 = 15.34, P < 0.001) but did not differ from the C-Tx group. There were no significant group differences in sex, age, duration of RA, number of comorbidities, or number and type of medication changes from T1 to T3. However, the control group had fewer participants receiving DMARDs compared with the C-Tx group at T1 (χ2 = 8.34, P < 0.02) and at T3 (χ2 = 6.06, P < 0.05). Participation in the exercise intervention was similar (both the C-Tx and H-Tx groups completed a median of 30 of the 36 exercise sessions) with similar ranges (4–36 for C-Tx, 3–36 for H-Tx).
Electrolytes, blood chemistries, hemoglobins, and hematocrits, obtained only to screen for other health problems, were within normal limits or remained stable for all participants at all assessments. Thyroid-stimulating hormone was measured at T1 to rule out hypothyroidism, which could affect symptoms. There were no significant group differences in number of participants needing thyroid replacement medication (χ2 = 0.202, P = 0.90) and no significant differences between cases and noncases on study outcomes. Therefore, changes in rheumatology medications or other health problems were unlikely to affect study outcomes.
For the first hypothesis, the model in Figure 2 is adequate (CFI = 0.91). Both overall symptoms at T1 and the C-Tx/control group contrast were significant predictors of overall symptoms at T3. Specifically, after controlling for T1 overall symptoms, the C-Tx group was significantly different from the control group (β = −0.108, P < 0.04) on overall symptoms at T3. There were no statistically significant differences in overall symptoms for the H-Tx group compared with the control group.
Structural equation modeling testing of effects of class exercise group or home exercise group versus control group on fatigue, pain, and depression from baseline (T1) to postintervention (T3). T1 or T3 after each measure = time period measure obtained either at T1 or T3; Mafglob = Multidimensional Assessment of Fatigue Global Index Score; Poms-F = Profile of Mood States Fatigue subscale; McG-D = McGill Pain Questionnaire Description of Pain; McG-I = McGill Pain Questionnaire Intensity of Pain; CESD = Center for Epidemiologic Studies Depression Scale; Poms-D = Profile of Mood States Depression subscale; depress = depression. *P < 0.04; ** P < 0.001.
For the second hypothesis, a Bonferroni correction for 10 repeated-measures ANOVAs resulted in setting a P value ≤ 0.005 for statistical significance (Table 4). The second hypothesis was supported. There were significant interaction effects for walk time (F = 4.52, P < 0.005) and left grip strength (F = 3.84, P < 0.005). Significant time effects were found for right grip strength (F = 28.42, P < 0.005) and aerobic fitness (F = 27.84, P = 0.005). Although not statistically different, the C-Tx group improved their aerobic fitness level by 12%, the H-Tx group improved by 10%, and the control group improved by 7%. No detrimental effects of exercise on disease activity were found.
| Dependent measure/period | Exercise group | F values | ||||
|---|---|---|---|---|---|---|
| C-Tx | H-Tx | Control | Time | Group | Interaction of time × exercise group | |
| ||||||
| Aerobic fitness | ||||||
| T1 | 22.50 | 23.32 | 21.10 | 27.84† | 1.44 | 1.33 |
| T2 | NA | NA | NA | |||
| T3 | 25.09 | 24.58 | 22.71 | |||
| Group | 23.80 | 23.95 | 21.91 | |||
| Left grip strength | ||||||
| T1 | 117.40 | 134.65 | 134.84 | 20.95† | 0.783 | 3.84† |
| T2 | 131.44 | 142.23 | 139.63 | |||
| T3 | 138.82 | 144.67 | 138.07 | |||
| Group | 129.22 | 140.52 | 137.51 | |||
| Right grip strength | ||||||
| T1 | 121.37 | 130.94 | 133.38 | 28.42† | 0.292 | 1.27 |
| T2 | 134.89 | 141.39 | 141.09 | |||
| T3 | 141.84 | 144.81 | 142.95 | |||
| Group | 132.70 | 139.05 | 139.14 | |||
| Walk time | ||||||
| T1 | 10.04 | 9.64 | 9.36 | 0.338 | 0.280 | 4.52† |
| T2 | 9.57 | 9.09 | 10.08 | |||
| T3 | 9.33 | 9.40 | 9.97 | |||
| Group | 9.64 | 9.38 | 9.80 | |||
| Total joint count | ||||||
| T1 | 32.16 | 29.04 | 37.14 | 4.03 | 3.21 | 1.69 |
| T2 | 32.80 | 23.19 | 32.99 | |||
| T3 | 31.04 | 23.72 | 35.14 | |||
| Group | 32.00 | 25.32 | 35.09 | |||
| ESR | ||||||
| T1 | 32.47 | 23.68 | 27.59 | 1.22 | 2.59 | 1.51 |
| T2 | 30.78 | 24.19 | 29.68 | |||
| T3 | 31.99 | 29.90 | 26.79 | |||
| Group | 31.75 | 23.26 | 28.02 | |||
| CRP | ||||||
| T1 | 1.33 | 1.25 | 1.30 | 3.34 | 0.34 | 0.16 |
| T2 | 1.20 | 1.07 | 1.14 | |||
| T3 | 1.13 | .86 | 1.02 | |||
| Group | 1.22 | 1.06 | 1.15 | |||
For the research question, CART was used to determine the level of scores on all predictor variables that were most predictive of exercise minutes. This analysis identified an interaction of the Global Fatigue Index and the Exercise Benefits/Barriers Scale scores. Participants with a global fatigue score ≥31 exercised an average of 62 minutes per week, whereas those with a global fatigue score <31 exercised an average of 85 minutes each week. Among participants with low fatigue, those with a score <117 on the Exercise Benefits/Barriers Scale exercised an average 62 minutes weekly, whereas those with a score ≥117 exercised an average of 95 minutes weekly.
The C-Tx group experienced a significant reduction in overall symptoms after 12 weeks of exercise, whereas the H-Tx group did not, despite participation in similar numbers of exercise sessions. The self-reported heart rates provide a plausible explanation. For the last 6 weeks of the exercise intervention, 72% of the C-Tx participants were exercising at 60% of their MHR and 16% were exercising at 80% of their MHR. In contrast, only 45% of the H-Tx group met the 60% criterion and only 4% met the 80% criterion. Thus, the H-Tx group exercised at a lower intensity than the C-Tx group, despite participating in the same number of sessions, and this may have contributed to less symptom improvement. Other researchers report positive effects for any amount of exercise and exercise intensity when compared with a sedentary lifestyle (48). Additional studies need to explore this possible intensity dose effect of exercise on symptoms. Other factors, such as the interaction of the C-Tx group with the exercise instructor and class participants, may have influenced ratings of symptoms in this study.
Both intervention groups decreased their walk time and increased their left grip strength. These data support use of the videotape by the H-Tx group as effective for functional improvements and these improvements are consistent with an earlier RCT (11) and our pilot study findings (22). Increased grip strength is likely due to the strengthening exercises of the intervention, which included work with elastic exercise bands. Interpretation of the left grip strength effect is tempered by the fact that the C-Tx group had lower left grip strength at T1, although not a statistically significant effect. Right grip strength was nonsignificant for the interaction and may have occurred due to less margin for improvement in the dominant right hand. Functional improvements are important because slow gait and weak grip are significant predictors of functional decline among elderly women in general (5), and among persons with RA specifically (49).
Although improvement in aerobic fitness did not reach statistical significance, the C-Tx group had the highest percentage improvement (12%). Larger improvements might have emerged if more participants in the treatment groups actually exercised at their target heart rate or if the exercise intervention continued beyond 12 weeks (3). Other researchers (50) suggest that it may be helpful to advise persons with low levels of activity to perform moderate-intensity exercise using 2 or 3 short bouts during the day instead of 1 longer session. The fact that some of our participants did not exercise at the intended intensity, or for the entire 30 minutes of the aerobic phase, suggests this may be a reasonable recommendation.
Fatigue and perceived benefits/barriers to exercise affected the amount of exercise performed. Participants with lower fatigue participated in more minutes of exercise. However, those with low fatigue who also had higher (better) scores on perceptions of benefits/barriers engaged in the most exercise. Perceptions of benefits and barriers did not affect amount of exercise among those with high fatigue. While we found an interaction of perceived benefits/barriers with fatigue, others have found perceived benefits alone to be a predictor of exercise in patients with RA (4, 14). Our findings reinforce the need for health providers to educate patients with RA about the many benefits of exercise, how to overcome barriers to exercise, and ways to manage fatigue. Part of fatigue in patients with RA may be due to a deconditioned state and regular aerobic exercise may decrease the detrimental effects of fatigue.
Limitations of this study include self-reported data, the fact that not all measures had been tested for reliability and validity in patients with RA, and possible selection bias from the convenience sample. Furthermore, absence of data on the 62 participants who dropped out prior to T1 precluded comparing them with participants who completed the study. Therefore, our sample may be biased toward more highly motivated participants. Lastly, with respect to limitations, the control group may have been less ill because fewer were receiving DMARDs.
In summary, this RCT, which used a larger sample than any RCT in the Cochrane Review, supported the positive effects of C-Tx and H-Tx exercise on walk time and grip strength, and found that fatigue and perceived benefits/barriers influenced exercise participation. Furthermore, overall symptoms for fatigue, pain, and depression were positively influenced in this selective group of patients with RA. However, given the differences found between the C-Tx and H-Tx groups, future studies should test use of an exercise videotape in combination with some in-class sessions. Our results also confirmed findings of 2 RCT reviews of RA exercise (6, 13) that exercise does not increase disease activity.
Dr. Neuberger had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Neuberger, Aaronson, Embretson, Cagle.
Acquisition of data. Cagle, Loudon.
Analysis and interpretation of data. Neuberger, Aaronson, Gajewski, Embretson, Miller.
Manuscript preparation. Neuberger, Aaronson, Gajewski, Embretson, Cagle, Loudon, Miller.
Statistical analysis. Aaronson, Gajewski, Embretson, Miller.
We acknowledge Herbert B. Lindsley, MD, and David Wilson, MD, who served as medical consultants, and Marian Minor, PT, PhD, who served as an exercise consultant. We also acknowledge the following former students who contributed to various aspects of this study: Marian Jamison, Ruthellyn Hinton, Sandra David, Kathleen Thornton, Mary Turnbull, Robert Spaniol, and Pat Twenter. Lastly, we wish to thank the men and women who generously gave their time and energy to participate in this study.
2151-4658/asset/olalertbanner.jpg?v=1&s=da9f09e93391fe585bb580a8ae4d5883379ee950)
