Abstract
Background Self-efficacy has important implications for health and functioning in people with limited mobility. However, the influence of self-efficacy on mobility in adults who use wheelchairs has yet to be investigated.
Objective The study objective was to: (1) estimate the direct association between wheelchair use self-efficacy and life-space mobility and (2) investigate an indirect effect through wheelchair skills.
Design This was a cross-sectional study.
Methods Participants (N=124) were adults who use a wheelchair, live in the community, and were 50 years of age and older (X̅=59.67, range=50–84), with at least 6 months of experience with manual wheelchair use; 60% were men. The 20-item Life-Space Assessment, the 65-item Wheelchair Use Confidence Scale, and the 32-item Wheelchair Skills Test-Questionnaire were used to measure life-space mobility, self-efficacy, and wheelchair skills, respectively.
Results Self-efficacy had a statistically significant association with life-space mobility (nonstandardized regression coefficient=0.23, 95% confidence interval=0.07, 0.39) after controlling for sex, number of comorbidities, geographic location, and assistance with using a wheelchair. This model accounted for 37.1% of the life-space mobility variance, and the unique contribution of self-efficacy was 3.5%. The indirect effect through wheelchair skills was also statistically significant (point estimate=0.21, 95% bootstrapped confidence interval=0.05, 0.43) and accounted for 91.3% of the direct effect of self-efficacy on life-space mobility. This model accounted for 39.2% of the life-space mobility variance.
Limitations Causality could not be established because of the study design. The self-report nature of data from volunteers may be influenced by recall bias, social desirability, or both.
Conclusions Wheelchair use self-efficacy had both direct and indirect associations with life-space mobility after controlling for confounding variables. Interventions targeted toward improving self-efficacy may lead to improvements in life-space mobility.
Wheelchairs are reported to be the most important form of assistive technology for older adults with mobility limitations.1 However, much of the evidence on the mobility of manual wheelchair users is limited to people living in long-term-care settings.2–4 In these studies, mobility was assessed with measures of life-space, which has been defined as movement extending from within the home to beyond the town or geographic region.5 Predictive models have identified several statistically significant factors, including wheelchair skills,3,4 the need for a seating intervention,2 depression,4 environmental barriers,4 functional ability,3 and social support.3 Despite knowledge of these factors, they are specific to people in long-term care settings and account for less than half of the life-space mobility variance. Evidence on the mobility of adults who use wheelchairs and live in the community is lacking, and the influence of other important mobility variables has not yet been investigated.
Self-efficacy6 with wheelchair use is emerging as a new and potentially important clinical and research construct. It has been conceptualized as the belief that people have in their ability to use their wheelchairs in various situations.7 Because self-efficacy has been shown to have important implications for people with limited mobility, including older people with lower extremity functional limitations,8 amputation,9 and knee osteoarthritis,10 it may similarly influence the mobility of wheelchair users. Preliminary evidence indicates that 39.0% of older Canadians who use wheelchairs report reduced self-efficacy with wheelchair use11 and that it has important participation implications.12 Therefore, it seems plausible for the construct to have a similar effect on the life-space mobility of people who use manual wheelchairs.
In this study, we examined the direct and indirect effects of wheelchair use self-efficacy on life-space mobility. The International Classification of Functioning, Disability and Health13 framework was used to organize the study variables. The primary hypothesis was that self-efficacy, conceptualized as a body function,14 would have a statistically significant association with life-space mobility, conceptualized as an activity, when confounding variables (ie, environmental and personal contextual factors) were controlled for. A secondary analysis was done to examine a potential indirect effect of self-efficacy on life-space mobility through wheelchair skills.
Method
This study was a secondary examination of data from cross-sectional research investigating the association between wheelchair use self-efficacy and frequency of participation in social and personal activities.14
Participants and Recruitment
A convenience sample of adults, 50 years of age and older, from British Columbia and Quebec, Canada, who use manual wheelchairs and live in the community was included for study. Participants also had to use a wheelchair on a daily basis and have at least 6 months of wheelchair-use experience. People with a Mini-Mental State Examination score of less than or equal to 2315 or an acute illness were excluded.
Clinicians from a rehabilitation center and from community-based home care rehabilitation teams from 3 regional health authorities in British Columbia recruited volunteer participants, as did clinicians from 2 rehabilitation centers in Quebec. People who met the study inclusion criteria were given information about the study. Community and senior centers as well as disability advocacy groups were given study information. All people who volunteered to participate met at a convenient location with a trained researcher for data collection. All study participants provided written informed consent.
Measures
Dependent variable: activity domain.
Life-space mobility was measured with the Life-Space Assessment.16 The Life-Space Assessment is a 20-item questionnaire that assesses the frequency at which people moved in 5 areas during the preceding month: within the home, around the home, in the neighborhood, in town, and outside of town. In addition, independence in life-space mobility is assessed. Total scores range between 0 and 120, with higher scores indicating more life-space mobility. This measure has excellent test-retest reliability for both older adults living in the community (intraclass correlation coefficient [ICC]=.96, 95% confidence interval [CI]=.95, .97)16 and people who use power wheelchairs (ICC=.87, 95% CI=.69, .92).17 Baker et al16 also reported moderate correlations in the expected directions with measures of physical performance (Spearman correlation [rs]=.60), activities of daily living (rs=−.40), and depression (rs=−.41) in older people.
Independent variable of interest: body function domain.
Self-efficacy with manual wheelchair use was measured with the 65-item Wheelchair Use Confidence Scale.7 This measure assesses the strength of self-efficacy in 6 conceptual areas: maneuvering around the physical environment, performing activities, knowledge and problem solving, social situations, advocacy, and emotions. Items are rated using a 0 to 100 response scale. Mean scores are calculated, with higher scores indicating stronger self-efficacy. In a recent methodological study of adults 19 years of age and older (median age=50.0 years, interquartile range=31.0–60.0) who use wheelchairs and live in the community, the internal consistency reliability (Cronbach alpha) of the Wheelchair Use Confidence Scale measurements was .92, the 1-week retest ICC was .84 (95% bootstrapped CI=.70, .90), and the standard error of measurement was 5.9.7 That study also provided evidence in support of the validity of the measurements through hypothesizing associations with relevant outcomes, including wheelchair skills (rs=.52), activities of daily living (rs=.32), depression (rs=−.43), and life-space mobility (rs=.38).7
Potential confounding variables: health, personal, and environmental domains.
The sociodemographic information form collected health-related variables, such as diagnosis, and personal factor variables, such as age, sex, education, marital status, employment/volunteer status, and income. The form also gathered data on environmental factor variables related to the wheelchair and the physical environment. Wheelchair-related variables included years of experience, hours of daily use, formal wheelchair training, and assistance with using the wheelchair (ie, “yes” [mobility, transferring, setup] or “no assistance”), and the physical environment variable was geographic location (ie, British Columbia or Quebec) because of differences in climate.
The number of comorbidities was assessed with the 18-item Functional Comorbidity Index.18 People indicate whether they have (yes=1) or do not have (no=0) each of the 18 health conditions listed. Total scores range from 0 to 18, with higher scores indicating more comorbidities. The 11-item Seating Identification Tool19 was used to assess the need for a seating intervention. Total scores range from 0 to 15, with higher scores indicating a greater need for a wheelchair seating intervention. In a sample of older adults who used wheelchairs, the Seating Identification Tool was found to have good test-retest reliability (ICC=.83) and good interrater reliability (ICC=.83).19 Perceived social support was measured with the 6-item Interpersonal Support and Evaluation List.20 Total scores range from 0 to 18, with higher scores indicating more social support. The properties of this measure have been established in the general population, including older adults,20 and the construct validity of this measure was demonstrated with the Community Integration Measure (r=.42) 21 and the Sense of Support Scale (r=.78).22 Finally, the number of physical environment barriers in the home and community was gathered with the Home and Community Environment Instrument.23 Scores on the home subscale range from 0 to 10, and scores on the community subscale range from 0 to 5; higher scores indicate more barriers. In a test-retest study, the agreement for the home subscale was 89% (kappa=.66), and that for the community subscale was 75% (kappa=.47).23 For items referring to “walking areas” in the community subscale, the wording was replaced with “areas you go to” to prevent ambiguity.
Potential intervening variable: activity domain.
The wheelchair skills variable was assessed with the 32-item Wheelchair Skills Test-Questionnaire.24 People are asked if they are able to complete a specific skill using their wheelchairs. Ratings include pass, fail, or not applicable. Total scores (percentages) are derived by dividing the number of individual skills passed by the total number of applicable skills; higher scores indicate more wheelchair skills. This measure was shown to have a high correlation with the performance-based Wheelchair Skills Test version 4.1 (rs=.89).25
The Barthel Index26 was used to assess functional ability, and the Hospital Anxiety and Depression Scale27 was used to assess anxiety and depression symptoms; the data from these instruments were used for descriptive purposes only.
Analyses
Descriptive statistics are presented as frequencies and percentages and as means and standard deviations. Multiple regression analyses were used to test the study hypotheses. A sample size of 123 was calculated a priori with G*Power version 3.1 (available at http://www.psycho.uni-duesseldorf.de/abteilungen/aap/gpower3/) using a moderate effect size (f2=0.15), an alpha value of .05, and a power of .80 for regression modeling with a maximum of 11 independent variables entered into the model. All regression assumptions were tested.28
Direct effect of self-efficacy on life-space mobility.
To estimate the most valid and precise association between self-efficacy and life-space mobility, we followed the 3-stage modeling strategy of Kleinbaum and Klein.29(pp169–173) Stage 1 is variable specification, stage 2 is interaction assessment, and stage 3 is confounding and precision assessment.
Stage 1: The self-efficacy variable was specified for entry because it was the key independent variable of interest. Two interaction terms were also specified for entry. A geographic location × self-efficacy interaction term was examined to determine whether the association between self-efficacy and life-space mobility differs between people in British Columbia and people in Quebec. An age × self-efficacy interaction term was also evaluated because evidence illustrates that self-efficacy diminishes with aging6; therefore, the association between self-efficacy and life-space mobility may likewise differ by age. The wheelchair skills variable was specified for entry for exploratory analyses of a potential indirect effect of self-efficacy on life-space mobility.
Data for 17 potential confounding variables (Tab. 1) were collected. Only confounding variables with published and plausible evidence of a relationship with the dependent variable were considered. To reduce the number of potential confounders for entry, such that the total number of variables (ie, self-efficacy, the interaction terms, wheelchair skills, and confounders) included for modeling did not exceed 11, we included only continuous confounders for which the correlation with the dependent variable had a magnitude of at least .2530(p525) or categorical confounders (eg, sex) with a statistically significant (P<.05) difference (determined with t tests or a 1-way analysis of variance) in the life-space mobility variable. After variables for inclusion were specified, residuals were examined to detect outliers and violations of regression assumptions.
Descriptive Statistics and Correlations With and Mean Differences in Life-Space Mobilitya
Stage 2: For evaluation of the geographic location and age interaction terms, the self-efficacy variable and lower-order interacting variables (ie, geographic location and age) were first forced into the model, followed by the interaction terms.29(pp207–210) If an interaction term was statistically significant, it was retained in the model along with the lower-order variable to ensure that the model was hierarchically well formulated.29(p206) The resulting model at this stage was considered to be the crude model in the next modeling stage.
Stage 3: To assess for confounding, we compared the self-efficacy estimates for the crude model and the adjusted model (ie, the crude model with the inclusion of the potential confounding variables).28(p190) We considered a percentage change in the adjusted model nonstandardized self-efficacy estimate—that is, [(crude estimate−adjusted estimate)/crude estimate] × 100—that exceeded 10% to be indicative of confounding.31(p261),32 If confounding was present, then further analyses were performed to identify subsets of the confounding variables that provided equivalent control of confounding but a more precise self-efficacy estimate. The precision of each confounder subset was evaluated by examining the width of the 95% CI around the self-efficacy estimate. The model with the same control of confounding as the adjusted model and with the narrowest CI around the self-efficacy estimate was deemed to provide the most valid and precise association between self-efficacy and life-space mobility and used in subsequent analyses.
Indirect effect of self-efficacy on life-space mobility through wheelchair skills.
After establishing a valid and precise association between self-efficacy and life-space mobility, we conducted a secondary analysis to investigate wheelchair skills as a potential mediator. We used the product-of-coefficients approach in combination with bootstrapping methods to derive the point estimate of the potential indirect effect and 95% CI.33 The proportion of the direct effect accounted for by the indirect effect was calculated as [(path a)(path b)/path c] × 100, where paths a and b are the indirect paths and path c is the direct path. The direct and indirect paths are shown in the Figure.
Direct and indirect effects of self-efficacy on life-space mobility. (A) Direct effect (path c) of self-efficacy on life-space mobility when confounding variables were controlled for. (B) Indirect effect (paths a and b) of self-efficacy on life-space mobility when confounding variables were controlled for. Values along arrows indicate nonstandardized regression weights.
SPSS version 19.0 (SPSS Inc, Chicago, Illinois) and the INDIRECT macro33 were used for the analyses.
Role of the Funding Source
This work was supported by the Canadian Institutes of Health Research (Doctoral Scholarship to Dr Sakakibara and Operating Grant IAP-107848) and the Michael Smith Foundation for Health Research (Senior Scholar Award to Dr Eng).
Results
Participants (N=124) in this research were adults who lived in the community (mean age=59.67 years, SD=7.49), had a variety of diagnoses, and had many years of experience with using a wheelchair (X̅=22.31 years, SD=16.05). Seventy-four participants (59.7%) were from British Columbia, and 74 (59.7%) were men. Thirty-nine participants (31.5%) required some form of assistance with using their wheelchair (eg, mobility, transferring, setup), and 22 (17.7%) received training to use their wheelchairs outside of rehabilitation. The participants had a mean score of 1.98 (SD=1.69) for the need for a seating intervention variable. People in this study had few depression and anxiety symptoms, and the mean score on the Barthel Index was 14.4 of 20 (SD=2.8). The participants reported lower self-efficacy (X̅=78.38/100, SD=19.19) as well as less life-space mobility (X̅=46.99/120, SD=17.84) than younger people who used manual wheelchairs in another study.7 The characteristics of the participants are shown in Table 1.
Direct Effect of Self-Efficacy on Life-Space Mobility
Stage 1: variable specification.
Overall, 10 variables were specified for modeling (as shown in Tab. 1): 1 independent variable of interest (self-efficacy), 2 interaction terms (geographic location and age), 6 potential confounders (geographic location, sex, number of comorbidities, formal training with wheelchair use, assistance with wheelchair use, and employment/volunteer status), and 1 potential intervening variable (wheelchair skills).
Potential collinearity was noted by a correlation of .84 between the self-efficacy variable and the wheelchair skills variable, but both were retained because the variance inflation factor values were low (range=1.03–3.77), thereby indicating no need for corrective action. No other regression assumption was violated.
Stage 2: interaction assessment.
Neither the geographic location interaction term nor the age interaction term was statistically significant. Therefore, the crude model shown in Table 2 included only the self-efficacy variable. This model accounted for 22.2% of the life-space mobility variance, and the self-efficacy estimate was 0.45 (95% CI=0.30, 0.59).
Confounding Model for Establishing a Valid and Precise Direct Effect of Self-Efficacy on Life-Space Mobilitya
Stage 3: confounding and precision assessment.
When the 6 potential confounders were controlled for, the self-efficacy estimate in adjusted model A diminished to 0.19 (95% CI=0.02, 0.36), as shown in Table 2. These 6 variables confounded the self-efficacy estimate in the crude model by 57.8%—that is, [(0.45−0.19)/0.45] × 100.
The 4-variable subset that included sex, number of comorbidities, geographic location, and assistance with wheelchair use confounded the self-efficacy estimate in the crude model by 48.9%—that is, [(0.45−0.23)/0.45] × 100—as shown in adjusted model B in Table 2. This subset also provided the most precise self-efficacy estimate relative to all other subsets of confounders tested; therefore, it was deemed to provide a valid self-efficacy estimate with the greatest precision. This model accounted for 37.1% of the life-space mobility variance, with 3.5% of the variance (Tab. 3) being explained by the self-efficacy variable.
Independent Contribution of Self-Efficacy to Life-Space Mobility When Controlling for Confounding Variablesa
Indirect Effect of Self-Efficacy on Life-Space Mobility Through Wheelchair Skills
The indirect effect of self-efficacy on life-space mobility through wheelchair skills was statistically significant (point estimate=0.21, 95% bootstrapped CI=0.05, 0.43). The wheelchair skills variable accounted for 91.3% of the direct effect—that is, [(path a)(path b)/path c] × 100=(0.21/0.23) × 100. This model, shown in Table 4 and the Figure (part B), accounted for 39.2% of the life-space mobility variance.
Indirect Effect of Self-Efficacy on Life-Space Mobility Through Wheelchair Skillsa
Discussion
In the present study, evidence was found to support the hypothesis that self-efficacy would have a statistically significant association with life-space mobility after controlling for confounding variables in adults who use wheelchairs and live in the community. In the existing literature, more than half of the life-space mobility variance in people who use wheelchairs remains unexplained after controlling for confounding variables in adults who use wheelchairs and live in the community.3 However, those predictive models did not consider the self-efficacy construct. Therefore, examination of self-efficacy in addition to important predictors identified in other studies might lead to a more complete understanding of life-space mobility in adults who use wheelchairs.
The findings of the present study contribute to the growing body of evidence indicating a need for the development and testing of targeted efficacy-enhancing interventions for people who use manual wheelchairs.12,14 Given that the life-space mobility of the participants in the present study occurred mostly within the home and neighborhood, as indicated by the low mean Life-Space Assessment score, improvements in self-efficacy may result in more travel to the community and, thus, community involvement. Improved self-efficacy may also result in more frequent excursions outside the home, less need for personal assistance with life-space mobility, or both, as measured with the Life-Space Assessment.
In theory, self-efficacy is amenable to information sourced from performance accomplishments, vicarious learning, verbal persuasions, physiological and affective states, or a combination of these factors.6 Moreover, research has illustrated the value of targeted theoretically based interventions in enhancing self-efficacy in people with impaired mobility.34,35 Evidence from an experimental pilot study demonstrated the modifiable nature of the wheelchair use self-efficacy construct in older people who did not have any experience using a wheelchair.36 Given the evidence that the Wheelchair Skills Training Program24 may be administered with structures in the community,37 the program offers an efficient and cost-effective clinical approach for improving self-efficacy through performance accomplishments in people inexperienced with wheelchair use.
The participants in the present study had experience using wheelchairs; therefore, modifying their efficacy judgments may be more complex than doing so with people inexperienced with wheelchair use. Therefore, interventions combining performance accomplishments and other sources of efficacy-enhancing information may be superior to any element alone in improving self-efficacy. Furthermore, the emphasis of each component in the intervention may differ depending on the person's strength of self-efficacy relative to his or her ability to use a wheelchair. Interestingly, evidence has illustrated that 28.0% of people who use manual wheelchairs have discordant self-efficacy beliefs and wheelchair skill capacities, with the majority having high self-efficacy and low skill.11 Conversely, a low self-efficacy–high skill profile may be more common in older adults. According to Bandura,6 self-efficacy issues for older people center around misappraisals of declining ability that, in turn, negatively affect self-efficacy. These data suggest that age may interact with self-efficacy to influence wheelchair skills. Therefore, the strategies used to elicit optimal self-efficacy improvements may differ depending on the person, the sample, or both and may require both age- and situation-specific approaches. Research investigating these strategies is needed.
Our secondary investigation revealed that self-efficacy has an indirect effect on life-space mobility through wheelchair skills. The results suggest that the association between self-efficacy and life-space mobility is almost entirely explained by wheelchair skills. Therefore, higher self-efficacy may lead to better use of wheelchair skills which, in turn, may increase life-space mobility. Although causality cannot be determined because of the cross-sectional nature of the study, social cognitive theory corroborates our results by postulating that self-efficacy has both direct and indirect effects on outcomes.6 In the context of the present study, self-efficacy may have contributed to the development of better ability because people with high self-efficacy were more likely to view ability-related impediments as surmountable and would have persevered when faced with difficulties to acquire better ability.6
Although our results are consistent only with the interpretation that wheelchair skills may mediate the association of self-efficacy with life-space mobility, findings from a previous experimental study36 have suggested that changes in self-efficacy may occur before wheelchair skills are acquired. In that experimental pilot trial, two 1-hour wheelchair skills training sessions led to statistically significant improvements in self-efficacy but not in wheelchair skills, as measured with the performance-based Wheelchair Skills Test.38 These findings, contrary to much evidence illustrating the value of the Wheelchair Skills Training Program at improving wheelchair skills,37–39 were attributed to the fewer number of hours of skills training administered (2 hours) relative to those administered in the other investigations (3–9 hours). These findings led to the speculation that skills training enhances efficacy before improving ability. Further research on a possible causal path between self-efficacy, wheelchair skills, and life-space mobility, as hypothesized in the present study, is warranted.
Limitations
The present study had limitations. In addition to the limitation of the study design regarding causality, the correlation between self-efficacy and the ability to use a wheelchair may have introduced collinearity into the secondary analysis investigating the indirect effect. In addition, the self-report nature of data from volunteers may be influenced by recall bias, social desirability, or both. The latter may be especially true for the use of the questionnaire version of the Wheelchair Skills Test. As a result, the data may overestimate a person's ability to use a wheelchair.
In conclusion, wheelchair use self-efficacy was found to have a statistically significant direct association with life-space mobility when confounding variables were controlled for. Interventions targeted toward improving self-efficacy may lead to improvements in the life-space mobility of adults who live in the community and use manual wheelchairs. The indirect effect of self-efficacy on life-space mobility was also statistically significant and was consistent with the interpretation that wheelchair skills may mediate the association of self-efficacy with life-space mobility. Longitudinal research is needed to test this hypothesized indirect effect.
The Bottom Line
What do we already know about this topic?
Thirty-nine percent of wheelchair-users, aged 50 and older report low wheelchair use self-efficacy (a person's belief in his or her ability to use a wheelchair in various situations). Evidence also suggests that this form of self-efficacy is associated with participation frequency. This association is mediated by life-space mobility (movement extending from within a person's home to beyond his or her town or geographic region) and participation restrictions, and the association is modifiable via wheelchair skills training.
What new information does this study offer?
This study provides evidence that self-efficacy has a positive and statistically significant association with life-space mobility, after controlling for confounding variables. The results also suggest a plausible mediating effect of wheelchair skills.
If you're a patient or a caregiver, what might these findings mean for you?
Wheelchair-users with high self-efficacy may acquire wheelchair skills more easily than those with low self-efficacy, which in turn may lead to more life-space mobility.
Footnotes
All authors provided concept/idea/research design and consultation (including review of manuscript before submission). Dr Sakakibara, Dr Miller, Dr Eng, and Dr Backman provided writing. Dr Sakakibara and Dr Routhier provided data collection. Dr Sakakibara, Dr Miller, and Dr Backman provided data analysis. Dr Sakakibara and Dr Miller provided project management. Dr Sakakibara, Dr Miller, and Dr Routhier provided fund procurement. Dr Routhier provided study participants. Dr Miller and Dr Routhier provided facilities/equipment and institutional liaisons. Dr Miller provided administrative support.
Ethics approval was received from all participating institutions.
This work was supported by the Canadian Institutes of Health Research (Doctoral Scholarship to Dr Sakakibara and Operating Grant IAP-107848) and the Michael Smith Foundation for Health Research (Senior Scholar Award to Dr Eng).
- Received March 17, 2014.
- Accepted June 2, 2014.
- © 2014 American Physical Therapy Association
References
Issue highlights
