Abstract
This study assessed the psychometric properties of the Arabic version of Exercise Self-Efficacy scale (ESE-A) among Jordanian outpatients with a variety of chronic diseases using descriptive cross-sectional design. Participants’ scores of ESE-A significantly correlated with their reported weekly exercise frequency (r = .23, p< .001), duration (r = .31, p< .001), and evaluation of their physical exercise (r = .39, p< .001). The construct validity was tested using exploratory factor analysis, which retained all items, and the scree plot showed one meaningful factor with an eigenvalue of 10.38 and an explained variance of 57.7%. Furthermore, Cronbach’s alpha was .89 and split-half coefficient was .83 indicating that the ESE-A is a reliable scale. The ESE-A was found to be a robust measure to evaluate exercise self-efficacy among Arabic patients with chronic diseases. Arabic researchers interested in exercise self-efficacy are invited to utilize the ESE-A in their studies to confirm its psychometric properties.
Introduction
According to the Jordanian Ministry of Health (2012), the number of Jordanian population suffering from chronic diseases is increasingly alarming. Hypertension was reported as the most prevalent disease (39%), followed by diabetes mellitus (DM; 29%) and cardiovascular diseases (10%; Department of Statistics [DOS], 2012), all of which are risk factors for other diseases such as end-stage renal disease (ESRD; The Hashemite Kingdom of Jordan Ministry of Health, 2009) and increased rates of mortality. For instance, the American Heart Association (2010) reported that cardiovascular diseases increase the mortality risk among patients with ESRD about 5 to 30 times than the normal people.
Due to lack of cure for the chronic diseases, self-care is the ultimate approach that can decrease their consequences and promote health care utilization (Darawad, Khalil, Hamdan-Mansour, & Nofal, 2016). An efficient self-care behavior is performing physical exercise, which can potentiate outcomes of patients’ treatment (Alramly, Darawad, & Khalil, 2013). Physical exercise has proven its efficiency in promoting functional status of patients with heart failure (Johansen, 2008); decreasing psychosocial stress among cardiac rehabilitation patients (Milani & Lavie, 2009); promoting physical, psychological, and social well-being among patients with multiple sclerosis (Dodd, Taylor, Denisenko, & Prasad, 2006); and decreasing severity and complications among patients with different chronic diseases (Chung, Cheng, Lin, Gau, & Chao, 2010; Saleh et al., 2015).
Even though the advantages of performing physical exercise for patients with chronic diseases are well-documented, their adherence to exercise recommendations is still less than the required (Hammad, Darawad, Haourani, & Demeh, 2015; Laoutaris et al., 2007). For instance, Darawad, Mosleh, et al. (2016) reported that patients with DM exercised for an average of only 2.9 hr weekly, which was less than the recommended for such group of patients.
A powerful measure to promote patients’ exercise behavior is their perceived exercise self-efficacy, which was defined by Bandura (2006) as patient’s confidence in his or her ability to perform exercise in the face of barriers. Exercise self-efficacy is part of the health promotion model (Pender, Murdaugh, & Parsons, 2002), which illustrates the causal relationship for promoting healthy behaviors. Many studies found a relationship between patients’ exercise self-efficacy and their commitment to exercise plan (Hagger, Chatzisarantis, & Biddle, 2001; Shin, Hur, Pender, Jang, & Kim, 2006) and with duration of exercise (Darawad, Mosleh, et al., 2016). Furthermore, exercise self-efficacy was among the variables that are recommended to be considered by researchers to identify the key antecedents of physical exercise among patients with chronic diseases, which can be used to classify those who need interventions to promote their exercise practice (Khalil & Abdalrahim, 2013).
To measure exercise self-efficacy, Bandura (1997) developed an 18-item Exercise Self-Efficacy scale (ESE). A single factor structure was revealed including all items with a Cronbach’s alpha of .89 demonstrating a high internal consistency. The English version was further utilized among Australian cardiac patients (Everett, Salamonson, & Davidson, 2009), and found to load on a single factor with high Cronbach’s alpha (.95). The ESE was shown to be valid and reliable when it was translated to other languages such as Dutch (Van der Heijden, Pouwer, & Pop, 2014), Korean (Shin, Jang, & Pender, 2001), and Persian (Noroozi et al., 2011). However, different factor structures were found among those studies, with an explained variance ranging from 77.5% to 96.4%.
To summarize, according to the literature, the ESE is a scale that is of great benefit to assess exercise self-efficacy among patients with different chronic diseases. It exists in many languages, and developing the Arabic version would be an important addition in this arena. Therefore, this study aims to assess the validity and reliability of the Arabic version of the ESE scale among Jordanians with chronic diseases, which is hoped to contribute to the body of knowledge and provide the Arabic health care professionals with a tool for their future research.
Method
Design
This validation study utilized a descriptive cross-sectional design and aimed to examine the psychometric properties of the Arabic version of Bandura’s Exercise Self-Efficacy scale (ESE-A).
Setting
Participants were selected from outpatient clinics (cardiac, endocrine, and renal) at four hospitals representative for the major health care sectors in Jordan: public, university-affiliated, and private. Those hospitals were randomly selected from a list of hospitals in each sector. The outpatient department in the selected hospitals receive a large number of patients (more than 60/day) with different conditions (mild-moderate) with appointments only. The care provided at those clinics is primary and secondary care including assessment of patient’s condition, prescribing medications, and admitting to hospital in severe cases (e.g., life threatening or cases that need continuous monitoring).
Sampling
A sample of Jordanian patients with selected chronic diseases was recruited using convenience sampling. To be eligible, participant had to (a) be 18 years old or more; (b) have a primary medical diagnosis of DM, hypertension, cardiovascular, joint, or renal disorders; (c) be able to comprehend Arabic language; and (d) agree to participate. These diseases are not inclusive of chronic diseases in Jordan, but are the most common. However, patients with severe mental or cognitive deterioration, those with physical disability, and those who were hemodynamically unstable were excluded. The hemodynamic status was established based on patient’s blood pressure and vital signs after consulting the nurses in the clinics and the initial assessment of data collector who were registered nurses.
Data Collection
Data collection started immediately after obtaining the ethical approvals from the Scientific Research Committee at the Faculty of Nursing, The University of Jordan, and the scientific committees in the selected hospitals. Research assistants visited the participating units and met the units’ heads to provide information about the study purpose and data collection procedure. Then, patients were screened for eligibility, and the eligible patients were invited to voluntarily participate in the study. Upon their agreement, participants were asked to read an information sheet that explained the purpose of the study and the participants’ rights. A private room was assigned for filling out the questionnaires, and participants were given the choice to take the questionnaire home and return it to the office of the unit head. Also, research assistants were available to answer participants’ questions concerning questionnaire items. Questionnaire administration took approximately 20 min. Data were collected between March and May, 2015.
Instruments
The instrument package of this study had three parts. The first part asked participants to report their demographic data including age, gender, marital status, level of education, and smoking status. In addition, participants were asked about primary medical diagnosis and comorbidities (out of five diseases: hypertension, diabetes, heart problems, renal problems, and joint problems). Primary medical diagnosis was confirmed by reviewing medical records. The second part asked participants regarding their exercise profile including frequency (exercise times per week), duration (exercise duration in hours per week), evaluation of their exercise activity, and health perception. Exercise was identified as any activity that the participant does regularly (2-3 times per week) for a minimum of 20 min including all activities (planned/unplanned, moderate/vigorous). This definition is consistent with the definition of exercise utilized in ESE questionnaire. For exercise evaluation, one question was utilized with four possible answers (1 = poor/none, 2 = little, 3 = good/moderate, 4 = very good/frequent). Finally, health perception was measured using a three-item scale (Mason-Hawkes & Holm, 1993) asking the participant to rate his or her own health (out of 4), rate health compared with others at same age (out of 3), and rate effect of health on exercise participation (out of 3). The total score is out of 10, with higher scores indicating greater health perception.
The third part measured exercise self-efficacy using the ESE-A, which asked participants to report their confidence in regularly doing 18 exercise behaviors on a 10-unit interval ranging from 0% “cannot do” to 100% “certainly can do” (Bandura, 2006). The average of the questionnaire items score was calculated, with the higher average demonstrating the patient’s greater confidence in his or her ability to perform exercise. The ESE has been utilized with different languages and shown to be valid and reliable among patients with various chronic diseases, including osteoporosis and osteoarthritis (Shin et al., 2006), DM (Noroozi et al., 2011), and cardiac rehabilitation (Everett et al., 2009), and among diverse population (Wilcox, Sharpe, Hutto, & Granner, 2005).
The ESE-A was produced by translating the original ESE questionnaire using the standard protocol of forward and backward translation. Initially, two of the researchers, including the principal, who are bilingual, translated the original ESE scale into Arabic. Then, a third researcher back-translated the items from Arabic to English. Later, two faculty members (colleagues) evaluated the equivalency of the meaning between the two versions, where minimal differences were found and the final wording of the items was agreed upon. The final version of the scale (ESE-A) was produced after consulting an expert in instrument development. Furthermore, the authors conducted a pilot study to ascertain that the instrument was clear and understandable for participants, and to evaluate the feasibility of the data collection process. The results of the pilot study revealed minimal language editing with no major changes.
Data Analysis
The Social Package for Social Sciences (SPSS-Version 21) was used to run all data analyses. The 18 items of the ESE scale were described in terms of descriptive statistics. Also, the normal distribution of the total ESE-A score was evaluated using Kolmogorov–Smirnov statistic. The sample size of this study allowed for analyzing the psychometrics of the ESE-A used in this study, and for comparing the results with the literature.
To estimate the internal consistency of the ESE-A scale, both Cronbach’s alpha and split-half coefficients were utilized. Then, exploratory factor analysis was used including the 18 items of the ESE-A scale and applying principal components and oblique promax rotation. The principal components analysis is a method used for data reduction, and oblique rotation methods assume that the factors are correlated (Gorsuch, 1983). Furthermore, Tabachnick and Fidell (2007) argued that requesting oblique rotation might be the best way to decide between orthogonal and oblique rotation. So, we have examined factor loadings using promax rotation (one type of oblique rotations) to see if these cross-loadings still appear. This actually occurred after confirming the intercorrelation between the items of the scale.
In factor analysis, an item loading of more than .40 was considered as the cutoff point for significant loading on the factor, and a value of .90 was utilized for inter-item correlation to prevent item redundancy (Nunnally & Bernstein, 1994). Also, a scree plot was utilized to observe natural breaks in factors. To be considered as a factor, a minimum of 10% of the common variance had to be explained. Finally, the criterion-related validity of the ESE-A was tested using Pearson correlation of the mean score with participants’ reported weekly exercise duration and frequency, whereas convergent validity was tested using Pearson correlation of ESE-A mean score with selected participants’ demographics including age, comorbidity, and health perception. In correlation analysis, a level of significance less than .05 was considered statistically significant.
Results
Sample Characteristics
A total of 350 participants met the eligibility criteria and were invited to participate in the study. However, only 272 (78%) participants agreed to participate. As seen in Table 1, participants’ mean age was 52.9 years old (SD = 12.1). The majority of the sample were females (51.8%, n = 141), married (78.3%, n = 213), and educated with less than high school (54.4%, n = 148). Out of five diseases, most of the participants reported having more than one disease (M = 1.66, SD = 0.89), and the most common medical diagnosis was hypertension (55.9%, n = 152), followed by DM (45.6%, n = 124). When asked about their exercise behavior, most of the participants evaluated their exercise as poor/none (43.8%, n = 119) or little (30.9%, n = 84), with an average of 1.97 times (SD = 1.81) of exercise per week for an average of 20.37 min (SD = 15.88) each time. Finally, participants’ mean score of perceived health was 6.5 out of 10 (SD = 1.75), and exercise self-efficacy was 29.5% (SD = 11.9). Table 1 provides a detailed description of participants’ demographics and exercise profile.
Description of Sample Demographic Variables (N = 272).
ESE-A Reliability and Validity
To assess the criterion-related validity, participants’ scores of ESE-A were tested for correlation with their reported weekly exercise frequency and duration, and evaluation of their physical exercise. Results (Table 2) showed significant positive correlation (r = .23, p< .001; r = .31, p< .001; r = .39, p< .001, respectively). Regarding the convergent validity, participants’ scores of ESE-A were found (Table 2) to have significant negative correlation with age (r = −.282, p< .001) and comorbidity (r = .−164, p< .001), and positive correlation with health perception (r = .450, p = .010). However, it is noted that many of these correlations were weak.
Correlation of ESE-A Score With Participants’ Demographics and Exercise Profile.
Note. ESE-A = Arabic version of Exercise Self-Efficacy Scale.
Significant at p< .01.
The construct validity of the ESE-A was tested using exploratory factor analysis, and utilizing the principal axis factoring and oblique promax rotation. The Kaiser–Meyer–Oklin measure of sampling adequacy was 0.934, which was in consistency with Bartlett’s Test of Sphericity (approximate chi-square = 3,731.707, p = .001), indicating sufficient sampling and sufficient correlation between items of the ESE-A, which allows performing factor analysis.
All items of ESE-A were retained in the factor analysis as each had a factor loading of more than .40 (Table 3). The item that had the highest loading (.847) was Item Number 5 “during or after experiencing personal problems,” whereas the item that had the lowest loading (.595) was Item Number 12 “when visitors are present.” Initially, the results of the factor analysis suggested a two-factor solution to the data. However, the scree plot showed one meaningful factor with an eigenvalue of 10.38 for the first factor and 1.33 for the second, with an explained variance of 57.7%. Therefore, a one-factor solution was most parsimonious. Furthermore, factors correlation matrix showed that Items 1, 4, 8, 12, and 13 are considered cross-loaded with correlation of .31 to .46, which means that these items are considered complex items whereas all other items are consider simple items.
Factor Analysis of ESE-A Scale.
Note. ESE-A = Arabic version of Exercise Self-Efficacy Scale.
The inter-item correlations were also evaluated (Table 4), which ranged from .281 to .845 with very minimal high interactions (above .70) indicating minimal possible item redundancy. The highest inter-item correlation was noted among Items 5 (during or after experiencing personal problems) and 6 (when I am feeling depressed), 5 and 7 (when I am feeling anxious), and 6 and 7 (r = .763, r = .738, r = .845, respectively). Similarly, the inter-item correlation was noted between Item 1 (when I am feeling tired) and Item 2 (when I am feeling under pressure from work; r = .809). Another high inter-item correlation was noted among Items 10 (after a vacation) and 16 (during a vacation; r = .783).
Inter-Item Correlation Matrix.
SE: SelfEfficacy.
Bolded values represnt the highest and lowest correlations.
The Cronbach’s alpha was utilized to evaluate the ESE-A internal consistency reliability, which should range from 0 to 1 and a value greater than .70 is considered acceptable (Polit & Beck, 2014). Along with Cronbach’s alpha, split-half coefficients for the ESE-A scale was estimated to determine its internal consistency. Results revealed a value of .89 for Cronbach’s alpha and .83 for split-half coefficients indicating that the ESE-A is a reliable scale. Furthermore, corrected item-total correlation (not including item itself within the total scale score) was estimated to assess the dimensionality of the ESE-A, which should achieve a value greater than .40 (Ware & Gandek, 1998). The analysis showed that ESE has a mean inter-item correlation of .54 ranging from .28 to .85.
Discussion
This study aimed to assess the psychometric properties of the ESE-A. Participants were patients with at least one chronic disease recruited from outpatient departments of the largest hospitals in Jordan. The diversity of participants and chronic diseases enhances the generalizability and the usability of the ESE-A scale among the Arabic population in the future studies. The internal consistency of the ESE-A, measured by Cronbach’s alpha and split-half coefficients, was supported with acceptable findings suggesting that the ESE-A is a reliable measure of self-efficacy for regular exercise among Arabic patients having a range of chronic diseases. Construct validity of the ESE-A was also supported by the results of the factor analysis, which revealed a one-factor model that explained 57.7% of the variance. This result was consistent with the single factor structure reported by many previous studies among patients with chronic diseases (Everett et al., 2009; Shin et al., 2001) and among older adults (Lee et al., 2009). The presence of some cross-loaded items indicate that although the pattern of loadings is strong overall, there is some complexity for which it is recommended that these items be reworded to fit in. However, further testing of the ESE-A among participants from different Arabic countries would add to the psychometric properties of this measure.
The ESE-A mean score was found to correlate with participants’ scores of reported weekly exercise frequency and duration, where participants who exercised more frequently and for a longer time had higher scores of exercise self-efficacy. In addition, participants’ scores on ESE-A significantly correlated with their age, comorbidities, and health perception. Patients who were younger, with less comorbidities, and with better health perception had higher scores of exercise self-efficacy. However, it is noted that in terms of magnitude, those significant correlations were weak. Similar correlations were noted in previous studies (Lee et al., 2009; Wilcox et al., 2005). This paradoxical result may be explained by the complex nature of the physical activity behavior, which is affected by many factors (Lee et al., 2009), particularly in a complex circumstance of patients with chronic diseases represented by the sample utilized in this study. Despite the weak correlations, it is acceptable to conclude that the results support the utilization of the ESE-A as a measure of exercise self-efficacy. Even though these findings are consistent with previous studies in that the more active people score higher on ESE scales (Everett et al., 2009), the clinical value of this decision needs to be validated with an objective measurement of the actual physical exercise practice.
Another advantage for the ESE-A is the very minimal item redundancy using .70 as a cutoff point for the inter-item correlation. It was noteworthy that the highest inter-item correlation was noted among items experiencing personal problems, feeling depressed, and feeling anxious. It seems that patients could not differentiate between depression and anxiety as it may look the same for the lay person, and that participants tend to think of emotional distress whenever having personal problems. Therefore, the results of this study support the suggestion by Everett et al. (2009) who suggested to combine the three items into one item that measures ESE in case of emotional distress. Similarly, a high inter-item correlation was noted between items feeling tired and feeling under pressure from work, which may indicate that participants think of “work” to be attached to being “tired.” Therefore, the item feeling tired is suggested to be rephrased taking this note in consideration. Another high inter-item correlation was noted among items after a vacation and during a vacation. It seems that participants think of vacation as one unit, which suggests combining them into one item that measures ESE during and after vacation.
To our knowledge, this is the first study to assess the Arabic version of the ESE scale measuring exercise self-efficacy among Arabic patients with chronic diseases. The Arabic population with different chronic diseases have been found to report very low scores on exercise frequency and duration measurements (Darawad & Khalil, 2012; Hammad et al., 2015; Mosleh & Darawad, 2015). This is supported by the low score reported in this study (M = 29.5%), which is less than those reported in literature (Lee et al., 2009; Wilcox et al., 2005). Such a difference can be explained by the high age and comorbidities of the sample, the heterogeneous sample of patients with various chronic diseases, the minimal efficacy information received by Jordanian patients that can increase their efficacy expectation (Lee et al., 2009), and the lack of exercise programs, places, and equipment. Therefore, the ESE-A would be useful for such population to be used in studies that assess participants in programs aimed at increasing exercise participation, or identify psychosocial associations that could be the target of interventions to improve exercise self-efficacy.
Despite the many strengths of this study, it has many limitations where the lack of more objective physical measurement of physical exercise is the most important, which hindered the clinical usability of the ESE-A. Therefore, replicating this study with correlating the ESE-A scores with objective physical measures of exercise involvement is highly recommended. Replicating this study among Arabic patients other than Jordanians is also recommended to confirm the psychometric properties reported in this study. Another limitation is the inclusion of patients with a list of chronic diseases that is not inclusive to the diseases prevalent in Jordan. However, those diseases are the most common. Despite these limitations, the study still presented very beneficial information pertaining the field of scales that measure exercise self-efficacy worldwide.
Conclusion
This is the first study to provide a preliminary systematic evaluation of the validity and reliability of the Arabic version of the ESE scale measuring exercise self-efficacy among Arabic patients with chronic diseases. The results of this study showed acceptable levels of the validity and reliability, which indicates that the ESE-A is a robust measure to evaluate exercise self-efficacy among Arabic patients with chronic diseases. The scale would be useful in studies assessing correlation of exercise self-efficacy with various variables including psychosocial variables. Further future studies including the ESE-A would identify its further advantages and disadvantages both in the clinical and research arenas.
Footnotes
Acknowledgements
Sincere thanks to the participants and to the directors of nursing within the participating hospitals.
Authors’ Note
This study was conducted during the Sabbatical Leave of the primary researcher (Muhammad Darawad).
Author Contributions
Muhammad Darawad planned and designed the study, is responsible for manuscript preparation, and provided supervision as well as technical and material support. Ayman Hamdan-Mansour co-supervised data collection process, and helped in preparing the literature review section. Amani Khalil was responsible for the manuscript preparation and she provided supervision as well as technical and material support. Also, Amani was responsible for data collection procedure. Diana Arabiat participated in planning and designing the study, and helped in preparing the literature review section. Osama Samarkandi participated in data analysis and the interpretation, and reviewed the final manuscript. Mahmoud Alhussami participated in data analysis and the interpretation, and reviewed the final manuscript. All members critically reviewed the article and have approved the final version submitted for publication.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors acknowledge the University of Jordan for funding this study.