Exercise, self-efficacy, and mobility performance in overweight and obese older adults with knee osteoarthritis

Authors


Abstract

Objective

To examine changes in mobility-related self efficacy following exercise and dietary weight loss interventions in overweight and obese older adults with knee osteoarthritis (OA), and to determine if self efficacy and pain mediate the effects of the interventions on mobility task performance.

Methods

The Arthritis, Diet, and Activity Promotion Trial was an 18-month, single-blind, randomized, controlled trial comparing the effects of exercise alone, dietary weight loss alone, a combination of exercise plus dietary weight loss, and a healthy lifestyle control intervention in the treatment of 316 overweight or obese older adults with symptomatic knee OA. Participants completed measures of stair-climb time and 6-minute walk distance, self efficacy for completing each mobility task, and self-reported pain at baseline, 6 months, and 18 months during the trial.

Results

Mixed model analyses of covariance of baseline adjusted change in the outcomes demonstrated that the exercise + dietary weight loss intervention produced greater improvements in mobility-related self efficacy (P = 0.0035), stair climb (P = 0.0249) and 6-minute walk performance (P = 0.00031), and pain (P = 0.09) when compared with the healthy lifestyle control intervention. Mediation analyses revealed that self efficacy and pain served as partial mediators of the beneficial effect of exercise + dietary weight loss on stair-climb time.

Conclusion

Exercise + dietary weight loss results in improved mobility-related self efficacy; changes in these task-specific control beliefs and self-reported pain serve as independent partial mediators of the beneficial effect of exercise + dietary weight loss on stair-climb performance.

INTRODUCTION

Knee osteoarthritis (OA) is a chronic degenerative disease that affects approximately one-third of all older adults in the United States. The joint damage and pain accompanying knee OA are primary causes of activity restriction and physical disability (1) and have a profound impact on quality of life in the elderly (2). Obesity is a modifiable risk factor for the development and progression of knee OA (3). Epidemiologic evidence suggests that weight loss may prevent the incidence of knee OA and alleviate adverse symptoms accompanying the onset of the disease (4). Consequently, weight loss interventions are now advocated in the treatment of overweight or obese patients with OA of the knee (5–7).

With the growing recognition of the importance of weight loss interventions for arthritis patients, the recently completed Arthritis, Diet, and Activity Promotion Trial (ADAPT) examined the effects of exercise and dietary weight loss interventions, both separately and in combination, in the treatment of overweight or obese older adults with knee OA. Findings from ADAPT demonstrated that combining exercise and dietary weight loss resulted in significant improvements in self-reported measures of physical function and pain symptoms and performance measures of mobility (8). The current investigation addresses the influence of the interventions in ADAPT on changes in mobility-related self efficacy and examines whether changes in self efficacy and self-reported pain mediate and/or are independent predictors of improvement in performance-based measures of mobility.

Perceptions of personal capabilities are important cognitive mechanisms that influence the health status and physical functioning of patients with arthritis. For example, self-efficacy beliefs and knee pain are independent predictors of activity restriction among older adults with knee OA (9). Performance-related self-efficacy beliefs have also been shown to be prospectively related to functional decline among older adults with knee pain (10). Recent research with physical activity also suggests that changes in performance-related control beliefs possess particularly important implications for the effectiveness of exercise interventions in the treatment of knee OA. Notably, Rejeski and colleagues (11) demonstrated that exercise therapy resulted in significant improvements in self efficacy for the performance of functional tasks. Moreover, changes in these mobility-related control beliefs and knee pain were found to mediate the improvements in performance measures of mobility achieved with exercise therapy.

Taken collectively, these findings reinforce the position that changes in mobility-related self-efficacy beliefs and perceptions of relevant physical symptoms, such as pain, are determinants of the functional benefits accompanying exercise participation. Therefore, in the present investigation, we examined the effects of exercise and dietary weight loss interventions, both separately and in combination, on mobility-related self efficacy in the ADAPT trial. A secondary objective of this study was to examine whether self efficacy and pain mediated the beneficial effects of the exercise and dietary weight loss interventions on performance measures of mobility.

MATERIALS AND METHODS

Design.

Complete details of the ADAPT design and methodology have been reported elsewhere (8, 12, 13). In brief, ADAPT was a single-blind, 18-month, randomized, controlled trial examining the effects of 4 interventions on various measures of physical function: exercise alone, dietary weight loss alone, exercise in combination with dietary weight loss, or healthy lifestyle control. ADAPT was conducted at the Claude D. Pepper Older Americans Independence Center of Wake Forest University, Winston-Salem, NC. All participants provided written informed consent that had received approval of the university institutional review board prior to participation.

Participants.

The eligibility criteria for participation in the study were age >60 years; calculated body mass index ≥28 kg/m2; self-reported knee pain on most days of the month; sedentary activity pattern with <20 minutes of formal exercise per week during the past 6 months; self-reported difficulty with at least 1 of the following activities due to knee pain: walking 0.25 miles (3–4 city blocks), climbing stairs, bending, stooping, kneeling, shopping, house cleaning, getting in or out of bed, standing up from a chair, lifting and carrying groceries, or getting in or out of a bathtub; radiographic evidence of tibiofemoral OA as determined by a single observer on the basis of weight-bearing anteroposterior radiographs; and willingness to undergo testing and intervention procedures.

Exclusion criteria included a serious medical condition that precluded safe participation in an exercise program such as coronary artery disease, severe hypertension, peripheral vascular disease, stroke, congestive heart failure, chronic obstructive pulmonary disease, insulin-dependent diabetes, psychiatric disease, renal disease, liver disease, active cancer other than skin cancer, and anemia; a Mini-Mental score <24; inability to complete the 18-month study or unlikely to be compliant; inability to walk without a cane or other assistive device; participation in another research study; excessive alcohol consumption of ≥14 drinks per week; or inability to complete the trial protocol, in the opinion of the clinical staff, because of frailty, illness, or other reasons.

A total of 2,209 older adults were prescreened via telephone interviews. Of this population, 1,596 individuals did not meet 1 or more of the eligibility criteria and another 297 refused to be contacted any further. A total of 316 participants were randomized into the study with the following group assignments: 82 in dietary weight loss alone, 80 in exercise alone, 76 in the combination exercise and dietary weight loss, and 78 in the healthy lifestyle control.

Measures.

Measures of mobility.

Participants completed 2 performance-based mobility tasks: a 6-minute walk task and timed stair-climb task. The 6-minute walk task was conducted in a gymnasium measuring 70 feet by 88 feet in area. Each individual was instructed to walk as far as possible in 6 minutes. Participants began walking at the command “go” and continued walking until they received the command “stop.” Participants were not allowed to carry a watch and were not provided any feedback during the test. Performance was measured as the total distance covered in feet. The stair-climb task involved ascending a set of 5 stairs, turning around on the top of the platform, and then descending. Performance was measured as the total time (in seconds) necessary to complete the task. These performance-based tasks have been shown to be valid and reliable tests of physical function in older adults with knee OA in previous research (9).

Walking self efficacy.

Prior to performing the 6-minute walk task, participants were asked to rate their confidence in their ability to walk around the gymnasium 2 times without stopping. This measurement was subsequently repeated for 5 additional levels of difficulty for the anticipated distances of completing 4 laps, 6 laps, 8 laps, 10 laps, and 12 laps without stopping. For each level of difficulty, participants were presented with a confidence ladder with 10 steps ranging from 0 (completely uncertain) to 10 (completely certain). Walking self-efficacy scores were calculated by summing the participant's confidence ratings across the 6 levels of difficulty and multiplying this result by 2 to produce a score ranging from 0 to 100. This hierarchical measurement protocol is consistent with the protocol developed by Bandura (14), and the walking self-efficacy scale has demonstrated adequate psychometric properties in several previous investigations (9, 11, 15).

Stair-climbing self efficacy.

Stair-climbing self efficacy was assessed using the same hierarchical measurement approach used for walking self efficacy. Specifically, prior to completing the stair-climb task, participants were asked to rate their level of certainty on the 0–10 confidence ladder that they could complete the task 2 times, 4 times, 6 times, 8 times, and 10 times without stopping. Stair-climb self-efficacy scores were also calculated by summing across the 5 levels of difficulty and multiplying this result by 2, yielding a score ranging from 0 to 100.

Pain.

Pain was assessed with the pain subscale of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) (16). Participants were asked to indicate the pain severity they had experienced during the past 48 hours due to knee OA on a scale ranging from 0 (none) to 4 (extreme). The WOMAC pain subscale consists of 5 items and total scores range from 0 to 20, with higher scores indicating greater pain.

Procedures.

Following recruitment, participants completed a series of clinic visits to verify their eligibility for the study. The baseline assessments were obtained during the screening visits and eligible participants were assigned, using a variable-block randomization procedure, into 1 of the 4 treatment arms: exercise alone, dietary weight loss alone, exercise and dietary weight loss, or healthy lifestyle control. Assessments of mobility-related self efficacy and physical function were obtained at data collection visits conducted at baseline and at 6 and 18 months after randomization.

Interventions.

Exercise.

Each participant assigned to the exercise alone or exercise and dietary weight loss treatment arms participated in three 60-minute exercise sessions per week. Specifically, each session comprised an aerobic phase (15 minutes), a resistance training phase (15 minutes), a second aerobic phase (15 minutes), and a cool down (15 minutes). The first 4 months of the 18-month intervention were facility based. At any time after the initial 4 months, participants who wished to exercise at home underwent a 2-month transition phase in which the participants alternated between exercising at the facility and exercising at their home. Hence, some participants remained in the facility-based program, others chose the home-based program, and some participants opted to engage in a combined facility/home-based program. At the beginning of a home-based phase, exercise leaders visited homes to work with participants in tailoring their individualized exercise regimens that were consistent with the study protocol.

The aerobic exercise involved walking within a heart rate range of 50–75% of heart rate reserve. The resistance training consisted of 2 sets of 12 repetitions of the following exercises: leg extension, leg curl, heel raise, and step up. Cuff weights and weighted vests were used to provide the resistance and a 1–1.5-minute rest interval was maintained between each exercise. Additional home visits or facility-based booster sessions were scheduled to assist participants who were having difficulty complying with the home-based exercise intervention. Attendance logs for center-based exercise and self-reported exercise logs for home-based sessions were used to gather adherence data and monitor progress. Exercise adherence was defined as the number of exercise sessions completed divided by the total number of prescribed sessions.

Dietary weight loss.

The dietary intervention was designed to produce and maintain an average weight loss of 5% across the 18-month intervention. The intervention was based on principles from the group dynamics literature (17) and social-cognitive theory (14). The major emphasis of the intensive phase was to heighten awareness of the importance of and need to change eating habits to lower caloric intake. Behavior change was facilitated through the use of self-regulatory skills including self monitoring, goal setting, cognitive restructuring, problem solving, and environmental management. One introductory individual session was followed by 16 weekly sessions consisting of 3 group sessions and 1 individual session each month. The transition phase included a session every other week for a total of 8 weeks. The goals for this phase included assisting participants who had not reached their weight loss goals in establishing new goals, and maintaining and preventing relapse in those participants who had reached their weight loss goals. The maintenance phase included monthly meetings and phone contacts, alternated every 2 weeks.

Exercise + dietary weight loss.

The exercise + dietary weight loss intervention involved completing the procedures previously described for both the exercise alone and the dietary weight loss alone programs. Delivery of the combined intervention was provided to participants consecutively on the same day and at the same location.

Healthy lifestyle control.

The healthy lifestyle control intervention served as a usual care comparison group with the 3 treatment arms and was designed to provide attention, social interaction, and health education. The group met monthly for 1 hour during the first 3 months. A health educator, who scheduled videotaped presentations and physician discussions on topics concerning OA, obesity, and exercise, organized the healthy lifestyle program. Monthly phone contacts were maintained during months 4–6 and bimonthly phone contacts during months 7–18. Compliance was defined as the number of sessions attended divided by the total number of sessions offered.

Statistical analysis.

The effects of the diet and/or exercise interventions on changes in self efficacy for walking and stair climbing at the 6-month and 18-month assessments were analyzed using mixed-model analyses of covariance (ANCOVA). Time (2 levels) and group (4 levels) were included as factors. We used the logarithm of stair-climb time and the Arcsin of the square root of self efficacy/100 as variance stabilizing transformations (18). All analyses were conducted using PROC Mixed in the SAS software version 8 (SAS Institute, Cary, NC), a procedure that analyzes all available followup information by providing maximum likelihood estimates of missing data. In each of these models, change from baseline in self efficacy was used as the outcome and age, sex, and the baseline self-efficacy value were included as covariates. The Tukey-Kramer adjustment for multiple comparisons was used when examining pair-wise differences. We examined intervention by time interactions and found that none were significant (all P > 0.35); therefore, they were not included in the analyses. We calculated the adjusted means for treatment averaging over time. Participants who completed baseline and at least one followup assessment were included in the analyses. Additionally, all analyses were conducted using the intent-to-treat principle. Following the primary analyses that tested the effects of the interventions on change in mobility-related self efficacy, secondary analyses were conducted to determine if self efficacy and pain mediated the effects of interventions on 6-minute walk and stair-climb performance. Specifically, a series of separate mixed-model ANCOVA models were tested to examine the effects of the interventions on change in self-reported pain and 6-minute walk and stair-climb performance; the effects of the potential mediators (self efficacy for walking and stair climbing and self-reported pain) on 6-minute walk and stair-climb performance; and a composite model testing the effects of the treatments on 6-minute walk and stair-climb performance while controlling for self efficacy for mobility and self-reported pain.

RESULTS

Participant attrition and adherence.

Of the 316 participants randomized into the trial, 252 (80%) completed the trial. As reported previously (8), retention of participants did not differ among the 4 intervention groups. Adherence was also not significantly different among the groups, with rates of 75% in the healthy lifestyle control, 72% in the dietary weight loss, 60% in the exercise only, and 64% in the combined intervention.

Effects of the interventions on self efficacy for mobility.

The main analyses in this study were conducted to examine the effect of the exercise and dietary weight loss interventions on change in the measures of self efficacy for mobility. Analysis of change in stair-climbing self efficacy yielded a significant main effect for treatment group (F[3,206] = 2.65, P = 0.05). Inspection of the group means provided in Table 1 demonstrates that participants in the exercise + dietary weight loss intervention significantly improved their stair-climb self efficacy compared with the healthy lifestyle control group. Analysis of change in walking self efficacy also revealed a significant main effect for treatment group (F[3,210] = 6.04, P = 0.0006). As illustrated in Table 1, participants in the exercise + dietary weight loss and exercise alone interventions significantly improved their walking self efficacy compared with the healthy lifestyle control group.

Table 1. Mean ± SD self-efficacy scores by task and treatment condition
Treatment groupBaselineFollowupChangeAdjusted score at followup
Self-efficacy scores for stair climb (range 0–100)    
 Health education70.24 ± 28.7572.28 ± 27.112.0483.20
 Diet only68.29 ± 27.9472.01 ± 27.123.7281.64
 Exercise only70.19 ± 26.2272.90 ± 26.022.7183.07
 Diet plus exercise63.84 ± 29.2577.25 ± 24.2813.4189.70
Self-efficacy scores for 6-minute walk (range 0–100)    
 Health education67.88 ± 31.9572.90 ± 31.895.0183.85
 Diet only68.91 ± 32.5374.03 ± 30.165.1183.79
 Exercise only67.49 ± 34.2182.03 ± 24.7114.5492.02
 Diet plus exercise66.44 ± 33.2684.95 ± 24.5918.5195.24

Although changes in mobility-related control beliefs are relevant outcomes of lifestyle interventions among older patients with knee OA, the minimal clinically significant difference (MCSD) associated with changes in mobility-related self efficacy has yet to be established. Accordingly, we estimated the MCSD for the mobility-related self-efficacy measures. To estimate the MCSD, we calculated the mean and SD of the transformed data for each self-efficacy outcome at baseline. We subsequently calculated small, medium, and large effect sizes. The following MCSD values represent the back transformed differences obtained by using half of the small, medium, and large effect sizes above and below the observed baseline mean: for walking self efficacy the effect sizes were 8.08, 20.04, and 31.60, whereas for stair-climb self efficacy the effect sizes were 6.70, 16.65, and 26.38, respectively.

Effects of the interventions on performance-related mobility and knee pain.

As reported in previous findings from the ADAPT trial (8), significant group main effects were observed for pain (F[2,232] = 3.37, P = 0.0193), 6-minute walk distance (F[3,196] = 12.75, P < 0.0001), and stair-climb time (F[3,212] = 3.18, P = 0.0249). Specifically, the exercise + dietary weight loss intervention resulted in significantly greater improvements in pain (t[232] = 2.35, adjusted P = 0.09) relative to the healthy lifestyle control group. In regard to the performance measures of mobility, the exercise + dietary weight loss intervention yielded a more favorable improvement in stair-climb time (t[212] = 2.85, adjusted P = 0.0249) relative to the healthy lifestyle control group, and both the exercise + dietary weight loss intervention (t[196] = −4.97, adjusted P < 0.0001) and the exercise alone intervention (t[196] = −4.55, adjusted P < 0.0001) resulted in significantly better changes in walking distance when compared with the healthy lifestyle control group (Table 2).

Table 2. Mean ± SD for mobility performance by task and treatment condition
Treatment groupBaselineFollowupChangeAdjusted score at followup
Stair-climb time    
 Health education9.44 ± 4.919.86 ± 5.560.418.60
 Diet only9.77 ± 5.709.86 ± 8.780.098.24
 Exercise only9.85 ± 4.569.15 ± 4.70−0.707.80
 Diet plus exercise10.38 ± 7.268.85 ± 5.35−1.537.54
6-minute walk distance    
 Health education1,422 ± 2691,411 ± 261−111,417
 Diet only1,406 ± 2541,433 ± 260271,447
 Exercise only1,417 ± 2511,551 ± 2971341,559
 Diet plus exercise1,360 ± 2801,524 ± 3161631,575

The independent and mediational role of self efficacy and pain on performance.

To evaluate the independent effects of self efficacy and pain on change in stair-climb time, composite mixed-model ANCOVAs were constructed using change in stair-climb time and 6-minute walk distance as outcome variables. In these models, covariates included baseline scores for each outcome (visit, age, sex, and treatment effect) whereas baseline status of pain or self efficacy, as well as change in pain or self efficacy, were considered for any independent effect that they had on change in each outcome. Results of these analyses revealed that baseline pain (F[1,191] = 7.60, P = 0.0064) and baseline self efficacy (F[1,191] = 7.03, P = 0.0087), together with change in pain (F[1,191] = 10.44, P = 0.0015) and change in self efficacy (F[1,191] = 35.93, P < 0.0001), were all independent predictors of stair-climb time beyond the treatment effect and other covariates mentioned above. In regard to the baseline predictors, these findings suggest that participants reporting less pain and higher self efficacy at baseline demonstrated superior stair-climb performance at followup. Additionally, the observation that change in each outcome was an independent predictor of performance suggests that participants demonstrating the greatest increase in self efficacy and decrease in pain exhibited the most favorable change in stair-climb performance. Similar results were found for 6-minute walk distance except that baseline pain was not statistically significant (F[1,178] = 22.27 for baseline self efficacy, P < 0.0001; F[1,178] = 9.05 for change in pain, P = 0.0030; F[1,178] = 28.47 for change in self efficacy, P < 0.0001).

Collectively, the primary results of this study and the findings reported previously by Messier et al (8) demonstrate that the ADAPT interventions had a significant effect on both the proposed mediators (self efficacy for mobility and knee pain) and the outcomes of interest (performance measures of mobility). Thus, we conducted secondary analyses in the present investigation to examine whether change in self efficacy and pain mediated the effects of the exercise and dietary weight loss interventions on change in the performance measures of mobility or whether there was evidence that change in self efficacy and pain had independent effects on the outcomes of interest. First, mixed-model ANCOVA analyses were conducted to test the effect of the proposed mediators on the measures of mobility task performance. Results of the model examining change in 6-minute walk performance demonstrated that baseline walking self efficacy (F[1,178] = 25.75, P < 0.0001), change in walking self efficacy (F[1,178] = 35.81, P < 0.0001), and change in pain (F[1,178] = 6.43, P = 0.0121) were significant predictors of change in baseline-adjusted 6-minute walk distance. Additionally, results of the model examining change in stair-climb performance demonstrated that baseline stair-climb self efficacy (F[1,192] = 7.96, P = 0.0053), change in stair-climb self efficacy (F[1,192] = 38.63, P < 0.0001), baseline pain (F[1,192] = 7.45, P = 0.0069), and change in pain (F[1,192] = 9.90, P = 0.0019) were significant predictors of change in baseline-adjusted stair-climb time. Thus, the findings from these models demonstrate that self efficacy and pain are significant independent predictors of change in walking and stair-climb performance.

Having established the required univariate relationships between the exercise and dietary interventions, self efficacy and pain, and the performance measures of mobility, we then tested a composite model examining the effect of the interventions on change in the measures of mobility performance after controlling for self efficacy and pain in the model. Results of the mixed-model ANCOVA analyses of stair-climb performance revealed that controlling for stair-climb self efficacy and pain in the original model attenuated, but did not eliminate, the significance of the treatment group main effect (F[3,192] = 2.78, P = 0.0425). Results of the ANCOVA analyses of walking performance revealed that controlling for walking self efficacy and pain in the original model did not reduce the significance of the treatment group main effect (F[3,178] = 11.46, P < 0.0001). However, in each analysis, the changes in self efficacy and pain remained independent predictors of mobility-related performance after including terms for the group effects.

DISCUSSION

In the ADAPT trial, the combination of exercise and dietary weight loss resulted in significant improvements in self-reported measures of physical functioning, pain, and performance measures of mobility. The present study examined the influence of the ADAPT intervention on changes in mobility-related self-efficacy beliefs. Moreover, we were interested in whether baseline levels or change in self-efficacy beliefs and pain were independent predictors of change in performance-based measures of function and/or whether change in self efficacy and/or pain mediated the effect that the interventions had on these indices of mobility disability.

Analyses revealed that both the exercise alone and the diet + exercise groups experienced significant increases in walking self efficacy. However, only the diet + exercise group reported significant improvements in self efficacy for stair climbing. Based on the estimated MCSD for the self-efficacy measures, the improvements in mobility-related self efficacy following the exercise alone and diet + exercise interventions represented small to moderate effect size differences. The differential effects of the exercise alone and combination interventions on change in self efficacy for 6-minute walk and stair-climb performance underscore the importance of specificity in the measurement of mobility-related self efficacy (14). Thus, the present findings suggest that although exercise alone increases self efficacy for walking, exercise must be combined with weight loss to influence efficacy beliefs involving the performance of more challenging mobility tasks such as stair climbing. Because the exercise only treatment did not differ from the healthy education control group on changes in stair-climb time in the ADAPT study (8), it appears that physical activity and weight loss should indeed be the intervention of choice in older, overweight, and obese adults with knee OA.

The results of this study underscore the importance of self-efficacy beliefs and pain in the treatment of knee OA, even within the context of lifestyle interventions. That is, both baseline levels and changes in self efficacy and knee pain were significant independent predictors of improvements in mobility performance. The only exception was that baseline levels of pain did not predict change in 6-minute walk times. These data are consistent with growing empirical evidence indicating that control beliefs and pain play key roles in the disablement process (10, 11, 19). Therefore, lifestyle interventions aimed at improving physical functioning in patients with knee OA should include components that target self-efficacy beliefs and pain management (19). In addition to maximizing effects on physical functioning, these types of intervention strategies have the added potential of enhancing adherence and the long-term maintenance of physical activity (15, 20).

In the present study, self efficacy and knee pain partially mediated the influence of the interventions on change in stair-climb performance. A previous study involving older adults with knee OA, the Fitness and Arthritis in Seniors Trial, found that self efficacy and pain completely mediated the effect of a physical activity intervention on stair-climb time (10). In the current study, the exercise only group did not differ from the health education control group on stair-climb time; thus, the effect in this study was driven by the combined exercise and dietary weight loss group.

In conclusion, the present study demonstrated that a combined dietary weight loss and physical activity intervention had unique effects on changes in self efficacy for a weight-dependent stair-climb task as compared with exercise alone. Additionally, both baseline values and changes in self efficacy and pain were significant predictors of improvement in mobility disability above and beyond the effects of the interventions. These findings add to the growing body of evidence demonstrating that control beliefs and physical symptoms are crucial to understanding and intervening the process of physical disablement (18). Furthermore, when treating obese, older adults who have compromised function due to knee OA, it is highly preferable to use a combined dietary weight loss and physical activity intervention as compared with a physical activity intervention alone.

Ancillary