Research emphasizes the negative impact of osteoarthritis (OA) on social participation, yet few studies have examined the roles of symptoms, activity limitations, and depression in this relationship. The present longitudinal study tested a model that hypothesizes that the relationship between physical symptoms and later participation restrictions among older adults with OA is mediated by activity limitations and depressive symptoms.
Participants were 184 community-dwelling senior adults (age ≥55 years) with a physician diagnosis of OA who were interviewed at 2 time points 18 months apart. Measures included demographic variables; a derived physical symptoms measure based on severity of pain, stiffness, and fatigue in the previous week; and depressive symptoms measured by the Center for Epidemiologic Studies Depression Scale. Measures of activity limitations and participation restrictions were derived by factor analysis of questions about difficulty in everyday life. Sequential multiple linear regression analyses controlling for demographic and illness-related variables were used to test for mediation.
Severity of time 1 physical symptoms was associated with difficulties in participation 18 months later. Sequential introduction of variables showed that this relationship was partially mediated by time 1 activity limitations and time 1 depressive symptoms. When both of these variables were included in the model, the effect of symptoms severity was completely mediated.
This study demonstrates the importance of taking into account both the physical (activity limitations) and psychological (depressive symptoms) consequences of OA symptoms and suggests that these factors act as a pathway to subsequent participation restrictions.
Osteoarthritis (OA) is one of the most prevalent chronic conditions among older adults, and is a significant predictor of disability and health care utilization (1–5). For example, studies indicate that pain associated with knee OA is an important predictor of disability, particularly for activities involving mobility, transfer, and instrumental activities of daily living (6). Other research on the impact of OA highlights the importance of examining a wide range of variables, including psychological factors in a range of social roles and activities, such as involvement in labor force and social and leisure activities (7–9). A prospective study of individuals with knee OA found that poor mental health status was a predictor of decreased physical function (8). Another study demonstrated that individuals with OA experienced more losses in the performance of leisure activities than controls (9). While this research points to a diverse group of variables that are important in understanding the impact of OA, an examination of disablement as a complex process in which several variables interrelate with one another to predict outcomes over time is lacking. The purpose of the present study was to test a model of OA disablement that hypothesized that the relationship between physical symptoms and later participation restrictions among older adults with OA is mediated both by activity limitations and depressive symptoms.
Theoretical models examining the interrelationships among health variables are not new and a number of different conceptual models of disablement exist (10–13). Many models share a similar structure and emphasize the inclusion of variables that represent the impact of a health condition at the level of the body (i.e., stiffness of the knee), the individual (i.e., difficulty in climbing stairs), and society (i.e., restriction of employment). For example, the International Classification of Functioning, Disability and Health (ICF) proposes that the consequences of disease should include measures of impairment, activity limitations, and participation restrictions (11). The ICF model suggests that there is a complex, sequential relationship among these components of the disablement process. Some research exists on these processes in other chronic conditions (14–17). However, few studies in arthritis have longitudinally examined these dimensions or tested the conceptual distinctiveness of the different constructs (18, 19).
Using the ICF as a conceptual framework to guide this research, we hypothesized that a mediational process exists, whereby physical symptoms severity at time 1 is not directly related to subsequent participation restrictions at time 2, but instead is associated with greater activity limitations and depressive symptoms at time 1 (Figure 1). That is, persons with OA who report greater pain, fatigue, and stiffness are more likely to report difficulty with activities and greater depression. In turn, activity limitations and depressive symptoms should be associated with later participation difficulties, such as restrictions in participating in community events, traveling, and visiting with family and friends.
PARTICIPANTS AND METHODS
Participants and recruitment
Participants were 184 community-dwelling senior adults age ≥55 years. All participants completed a face-to-face, interview-administered survey consisting of 2 questionnaires 18 months apart that examined OA and independence. Approximately two-thirds of the participants resided in a large urban center located in the Greater Toronto area of Ontario, Canada. The remaining participants resided in smaller urban or rural communities in the same province. Participants were recruited through community newspaper advertisements and posters in rheumatology offices and rehabilitation clinics. Individuals were first screened over the telephone to determine if they were eligible to participate. Eligibility criteria included 1) self-report of a physician diagnosis of OA; 2) difficulty attributable to OA in at least 1 activity such as personal care, household activities, or mobility; 3) no other comorbid condition causing disability (e.g., emphysema, stroke); and 4) fluency in English. A total of 248 individuals were eligible to participate in the study. The primary reason for ineligibility was the report of another comorbid condition resulting in disability. Of the 248 individuals interviewed at baseline (time 1), 184 (74%) were contacted and agreed to a followup interview 18 months later. Independent t-tests comparing age and severity of physical symptoms and chi-square tests comparing sex, education, and employment revealed no significant differences between respondents (n = 184) and nonrespondents (n = 64) at the followup interview. Participants were interviewed in their homes by trained interviewers using a closed-ended, structured questionnaire. Interviews lasted 90 minutes on average. Informed written consent was obtained from all participants.
Information was collected on respondents' age, sex, education, employment status, and living arrangements.
Respondents were asked about the number of years since diagnosis. They were also asked to provide details on the location of their arthritis by reporting up to 6 symptomatic joints. This information was aggregated into 2 variables, one related to a report of either a hip or knee affected by OA and another related to a report of a hand or wrist affected by the disease.
Severity of physical symptoms.
Three questions concerning arthritis symptom severity asked about pain, stiffness, and fatigue in the previous week. Research suggests that these symptoms, including fatigue, are among the most frequently reported by individuals with OA (20–22). Similar items have been used in other research and have revealed preliminary evidence for the construct validity of the questions, including associations of symptoms with disability at home and at work, changes in employment, use of assistive devices, coping behaviors, and negative disease appraisals (7, 23–26). Responses were rated on a 4-point Likert-type scale (where 1 = none, 2 = mild, 3 = moderate, and 4 = severe). For the purpose of this study, a combined symptom severity variable was created. The internal consistency of this measure, assessed by Cronbach's alpha, was 0.75.
The 20-item Center for Epidemiologic Studies Depression Scale (CES-D) was used to measure depressive symptoms (27, 28). Participants were asked to report on the frequency of symptoms experienced during the previous week. Responses were rated on a 4-point Likert-type scale, where 1 = rarely or none of the time (<1 day), 2 = some or a little of the time (1–2 days), 3 = occasionally or a moderate amount of time (3–4 days), and 4 = most or all of the time (5–7 days). Scores ≥16 are considered indicative of depressed mood. Cronbach's alpha, a measure of the internal consistency of the scale, was 0.87 in this sample. This is comparable with data from other community samples. The mean ± SD CES-D score was 11.3 ± 10.
Activity limitations and participation restrictions.
Thirteen items on activity limitations and participation restrictions were derived from a set of questions about difficulty with activities of everyday life. Questions on personal care, mobility, and household activities were drawn from similar items used on the Stanford Arthritis Center Health Assessment Questionnaire (29), the Multidimensional Functional Assessment Questionnaire (Osteoarthritis Research Society International) (30), and other scales commonly used to assess difficulties with activities of daily living. These activities were supplemented with additional social and leisure activities. Respondents were asked to think about performing the activity “without any help from another person or from a special gadget or piece of equipment.” Responses were rated on a 4-point Likert-type scale (where 0 = no difficulty, 1 = some difficulty, 2 = a lot of difficulty, and 3 = not able to do the activity).
Exploratory factor analysis was used to investigate whether the 13 items concerning difficulties with activities corresponded to distinct activity and participation domains. Principal components analysis was used as a method to extract factors. Orthogonal rotation was performed to assess factor loading. Criteria used to retain an item were 1) factor loading ≥0.60 and 2) loading on a single factor. Criteria used to identify factor(s) were 1) eigenvalues ≥1 and 2) a minimum of 3 items in each factor (31). Scores for each factor were obtained by summing the responses of the corresponding items. Cronbach's alpha was calculated to assess the internal reliability of the items composing the extracted factors.
To test our conceptual model, Pearson's correlation coefficients were initially calculated examining the relationships among time 1 physical symptoms, depressive symptoms, activity limitations, and participation restrictions. Then, sequential multiple linear regression analyses were carried out and standardized regression coefficients were calculated controlling for age, sex, education, employment status, living arrangements, illness duration, and hip/knee affected. Participation restrictions were assessed by modeling time 2 participation restrictions controlling for time 1 participation restriction scores.
Six regression models, deemed direct effect models, were initially estimated. The first 3 investigated whether severity of physical symptoms was associated with depressive symptoms, activity limitations, and later participation restrictions. Two other models investigated whether depressive symptoms were associated with activity limitations and subsequent participation restrictions. The last direct effect model investigated whether activity limitations were associated with later participation restrictions.
Further regression analysis was performed to test for mediation. To function as a mediator, a variable had to satisfy 3 conditions: variations in the predictor variable significantly accounted for the variations in the presumed mediator, variations in the presumed mediator significantly accounted for variations in the outcome variable, and the effect of the predictor on the outcome variable was diminished or was no longer significant if adjusted for the presumed mediator (32).
Based on these criteria, 4 additional regression models, deemed mediated models, were estimated. The first of these models investigated whether the effect of physical symptoms on activity limitations was mediated by depressive symptoms (M1). The other 3 models investigated whether the effect of physical symptoms on later participation restrictions was mediated by activity limitations (M2), depressive symptoms (M3), and both activity limitations and depressive symptoms (M4).
The extent of mediation was quantified by calculating the percentage change in the estimated effect (beta) for the predictor variable from the direct effect model compared with the estimate from the model that included the presumed mediator(s) (33). SPSS software, version 12.0 for Windows (SPSS, Chicago, IL) was used in all data analyses.
Sample characteristics are shown in Table 1. The mean age of the participants was ∼68 years and the sample was primarily female (84%). Approximately 40% of the participants lived alone. Overall, the sample was relatively well educated and the majority of participants were not in the paid labor force. There was a great deal of variation in the duration of OA, with the mean number of years reported as 11.6. Approximately 88% of participants reported that either their hip or knee was affected by the disease. Approximately two-thirds of the participants reported moderate or severe pain, stiffness, or fatigue associated with OA in the previous week.
Values are the number (percentage) unless otherwise indicated. CES-D = Center for Epidemiologic Studies Depression Scale.
Age, mean ± SD years (range 55–86)
68.3 ± 8.0
Less than secondary school
Some postsecondary school
Not in labor force
In labor force
Illness duration, mean ± SD years
11.6 ± 11.0
Hip or knee
Hand or wrist
Depression (CES-D score ≥16)
The results of the factor analysis are shown in Table 2. Three factors explained 58% of the total variance. Factor 1 comprised 5 items related to difficulties in social and leisure activities and was similar to the ICF concept participation restrictions. When summed, Cronbach's alpha, a measure of the internal consistency of the items, was 0.80 for time 1 and 0.81 for time 2. Factor 2 comprised 5 items related to difficulties in movement and transfer activities, and was labeled activity limitations (α = 0.80). A third scale related to upper limb activities was derived, which included only 3 items and was excluded from further analyses because of its unsatisfactory reliability (α = 0.62).
Table 2. Results of factor analysis of items related to difficulty with activities and percentage of participants with any difficulty in activities*
Any difficulty = some difficulty, a lot of difficulty, or not able to do the activity.
Taking a tub bath
Getting on and off the toilet
Standing up from a chair
Climbing up a flight of stairs
Getting in/out of car
Taking care of appearance
Eating a meal
Preparing a meal
Doing errands or shopping
Getting around in the community
Visiting with family and friends
Participating in community activities
Percentage of variance explained
Physical symptoms, depressive symptoms, activity limitations, and participation restrictions were significantly correlated (P < 0.001) with one another, fulfilling the first criterion to proceed in further tests for mediation. Physical symptoms was correlated with depressive symptoms (r = 0.48), activity limitations (r = 0.53), and participation restrictions (r = 0.55). Depressive symptoms was correlated with activity limitations (r = 0.36) and participation restrictions (r = 044). The highest correlation was between activity limitations and participation restrictions (r = 0.67).
Results from the direct effect models are shown in Figure 2A. After controlling for demographics and illness-related variables, the estimates (β) from the 6 models for the main independent variables of interest were all significant at P < 0.01, fulfilling another criterion for mediation.
Having fulfilled all prerequisites to test for mediation, final models including presumed mediators were tested and the results are shown in Figure 2B and Table 3. The results from model M1 show that, when accounting for depressive symptoms, the effect of time 1 physical symptoms on activity limitations was diminished from 0.50 to 0.42 (16%), suggesting partial mediation in this relationship. In model M2, after adjusting for activity limitations, the effect of physical symptoms on later participation restrictions diminished by 19%, indicating that this relationship was partially mediated by activity limitations. Model M3 showed that the effect of physical symptoms on subsequent participation restrictions was partially mediated by depressive symptoms as shown by a 29% reduction in the coefficient. Model M4 showed the results of controlling for both mediating factors (activity limitations and depressive symptoms): time 1 physical symptoms was no longer significantly associated with later participation restrictions (β = 0.12, P = 0.15), corresponding to a 43% decrease from the direct effect.
Table 3. Standardized regression coefficients (β) for variables in mediated models of participation restriction in older adults with osteoarthritis
In addition to the independent variables of interest in the conceptual model, age was significantly associated with subsequent participation restrictions in 3 of 4 mediated models, and having a hip or knee affected by OA was associated with activity limitations in 1 of the models.
This research used a conceptual model of disablement and longitudinal data to examine the largely ignored impact of OA on participation. The findings enhance our understanding of the consequences of OA on individuals' participation in social and leisure activities and emphasize that participation restrictions are a result of the complex interrelationships of variables that include not only symptoms of the disease but also disability in other areas of life (activity limitations) as well as psychological distress (depressive symptoms).
The results of factor analysis of items on difficulties in daily life activities suggested that the items related to movement and transfer and to more social activities were distinct dimensions that were similar to the concepts of activity limitations and participation restrictions as described in the ICF (34, 35). Our findings in this respect are similar to those of others (18). To date, there are few studies in arthritis that have assessed activity limitations and participation restrictions as measures of disability (19, 36, 37). Further research on OA disability is necessary to generate additional areas of participation items that may be affected by OA (e.g., employment) (25, 26) and to link the items to categories in the ICF framework, looking not only at the dimensionality of the items, but also at their relationship with contextual factors.
Our results emphasize the importance of looking at disablement in OA as a process and using conceptual models to examine the complexity of the relationship among variables that measure distinct dimensions of the consequences of this chronic illness. The finding that activity limitations and depressive symptoms mediated the relationship between physical symptoms and later participation restrictions allowed us to go beyond the results of regular regression analyses. Specifically, the findings of the current study suggest that symptom severity does not directly affect people's ability to participate in social and leisure activities. Instead, symptoms such as pain affect people's mood and ability to perform other types of tasks such as getting in and out of a car and standing up from a chair. Difficulties with these tasks and changes in mood, in turn, are related to social and leisure tasks. In part, these results may be due to the fact that difficulties with basic tasks are often part of more complex social and leisure activities. For example, visiting others and getting around the community may involve climbing stairs and standing up from a chair. The importance of mood in participation restrictions suggests that feelings of upset related to pain may spill over into other areas of life, leaving people with more difficulty engaging in other activities. Finally, people may also decide to prioritize their activities, as shown by other research (7). For example, difficulties with personal care tasks such as toileting and bathing may now require additional time and effort and leave people unwilling to try to engage in social or leisure activities. Future research should examine people's priorities and other ways of adaptation to disability in OA.
This study highlights the importance of addressing psychological variables such as depressive symptoms to understand the disablement process in OA. Previous research on the consequences of arthritis has found variability of relationships among depression, physical symptoms, and physical disability (38–40). This study demonstrated support for the hypothesis that changes in mood may be a result of symptoms, which then affect participation. However, further research is necessary and future studies should also examine the potential role of depressive symptoms as a predictor of subsequent worsening pain (41) and as an outcome of participation restrictions (42).
The finding that age was a predictor of participation restrictions is consistent with previous research showing age as a risk factor for disability in elderly cohorts (43, 44). This result suggests that age may have been a marker of comorbidities related to disability that were not captured by the eligibility criteria of this study. The association of having a hip or knee affected by OA with activity limitations is also consistent with previous research connecting lower limb OA to problems in physical mobility (45, 46). The measure of limitations in activities used in this study, based mainly on lower limb functions, may have contributed to our findings, which point to the broader implication of having mobility problems in OA.
There are several limitations to this research that need to be acknowledged. Our sample was a purposive one and there was a high proportion of white and highly educated women among the participants. Although the mean age and the high percentage of participants with difficulty in activities and main joints affected by arthritis made this sample comparable with other studies of community-dwelling older adults with OA, the study should be replicated with more diverse groups. It is also important for future research to conceptually clarify and distinguish activity limitations and participation restrictions. Our measure was derived from statistical analyses using questions commonly asked of persons with physical disabilities such as OA. However, other aspects of activity limitations (e.g., walking) and participation (e.g., employment, close relationships) should be addressed. Finally, our measure of symptom severity would benefit from additional efforts to assess its validity and reliability. Other research has noted the importance of fatigue in addition to pain and stiffness in OA (20–22). However, it is possible that the particular selection of symptoms assessed may alter the relationships among activity limitations, depression, and participation. This study should also be replicated in other musculoskeletal conditions, such as rheumatoid arthritis, where fatigue has been highly prevalent and found to be associated with depression (47).
With its limitations acknowledged, this study extends our understanding of the relationship between the components of the disablement process in OA. The findings highlight the importance of taking into account both the physical (activity limitations) and psychological (depressive symptoms) consequences of OA and suggest that these measures act as a pathway to explain the association between physical symptoms and individuals' engagement in society.
Dr. Machado 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. Machado, Gignac, Badley.
Acquisition of data. Gignac.
Analysis and interpretation of data. Machado, Gignac, Badley.