Introduction

Musculoskeletal disorders (MSDs) are often correlated with ergonomic risk factors and also socioeconomic characteristics of workers.1 Globally, biopsychosocial factors of the workplace affect the majority of the world’s population who spend most of their waking hours in their workplace.2 One of the most important risk factors for computer users in the work sites is prolonged sitting without doing stretching exercises (SEs).3 Work-related MSDs (WMSDs) are one of the prevalent health problems at the work sites.4 Repetitive motions, excessive inactivity or prolonged sitting as well as psychological stresses have been associated with WMSDs among computer operators.5 SEs can lead to permanent lengthening of ligaments and tendons6 and it seems to have an impact on decreasing WMSDs especially among computer operators.7 8

In a previous study it was argued that inactivity and not doing SEs were prevalent among Iranian computer operators.9 The health promotion model (HPM) is one of the comprehensive models that determine the influencing factors that affect health promoting behaviours especially at work sites. This model describes factors like perceived barrier/benefit to action, perceived self-efficacy, interpersonal influences, commitment to a plan of action, immediate competing demands/preferences and situational influence on health behaviour—for instance SEs—in the context of the work site.10 However, a previous study11 showed that other factors such as stimulus control, counterconditioning and self-regulation were influencing exercise behaviours. It has been documented that not doing exercise among Iranian office workers was prevalent and, on the other hand, there was no valid instrument to measure real needs of Iranian computer users based on HPM constructs to assess the causes for not doing Stretching Exercise (SE). A previous study revealed that the weight of the influencing factors on stretching training can vary depending on the cultural context.12 Therefore, developing a reliable instrument for assessing factors influencing SEs is essential to understanding and addressing the interventional programme to promote SE. In this context, the objective of this research was to develop and validate a culturally based instrument to evaluate factors influencing SE among a sample of Iranian computer users.

Objectives

The objective of this research was to develop and validate a culturally HPM-based instrument to evaluate SE influencing factors among a sample of Iranian computer users.

Methods

This cross-sectional study was part of a PhD thesis in Tarbiat Modares University, Tehran, Iran. All the participants signed an informed written consent form to participate in this study.

For this study, first of all, a questionnaire including 86 items pertaining to the mentioned constructs of HPM—in the context of WMSDs and based on the existing evidences—was designed. The validity of the instrument was determined by a sample of 420 office employees who were working at health centres and were eligible due to the inclusion/exclusion criteria. The inclusion criteria were having no disability or illnesses to prevent SEs and signing the written consent form. So, those suffering from any defect or illness interfering with SE were excluded from the study. Both quantitative and qualitative approaches were taken for face validity of the questionnaire. In the qualitative approach, 30 office employees assessed each item of the questionnaire for ‘ambiguity’, ‘relevancy’ and ‘difficulty’. In this process, three items needed to be improved.

For the quantitative approach, the same office employees were asked to determine the importance of each item through a 5-point Likert Scale. In this way the impact score for each item was calculated. As the impact score of 1.5 or above was satisfactory, all the items were approved for the instrument.

Content validity was done by both qualitative and quantitative methods. For the qualitative method an expert panel consisting of 15 specialists, including 6 health education specialists, 2 psychologists, 1 psychometric specialist, 1 physiotherapist, 1 neurological pain manager, 1 orthopaedic specialist, 1 physical medicine expert and 1 nurse with  experience on pain management, checked all the survey items. These experts inserted their recommendations into the questionnaire. Moreover, they also evaluated the questionnaire for ‘grammar’, ‘wording’, ‘item allocation’ and ‘scaling’ indices. This expert panel was asked to comment on item relevance, item comprehensiveness and any confusing meaning.

For quantitative content validity, the Content Validity Ratio (CVR) and Content Validity Index (CVI) were used. The necessity of an item was assessed through CVR and items with a score <0.4 were deleted according to Harrington.13 The simplicity, relevance and clarity of the items were assessed through CVI and a value of 0.79 or above was considered satisfactory for each item.

According to a rule of five individuals for each item (86×5), a sample size of 385 computer users was estimated for exploratory factor analysis (EFA). However, for greater accuracy the sample size was increased to 420 individuals.14 Multistage cluster sampling was applied to select the sample for psychometric evaluation of the instrument. First, from 10 health networks of Shahid Beheshti University of Medical Sciences, the North, Shemiranat and East networks were selected. Then eight health  centres were selected from each health network, and 150 computer users from each health centre in the North and Shemiranat networks and 120 office employees from the health centre in the East network were randomly selected. Figure 1 shows the sampling procedure.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow of the procedure for sampling office employees. Shahid Beheshti University of Medical Sciences (SBUMS).

The primary questionnaire included 19 demographic questions and 86 questions relevant to the 11 constructs of HPM and other evidences. Each construct included five to nine questions. The construct validity of the questionnaire was examined through EFA. Principal component analysis with varimax rotation was performed to extract the underlying factors. Factor loadings ≥0.5 were considered appropriate. Eigenvalues >1 and Scree plots were used for determining the number of statements. The Kaiser-Meyer-Olkin (KMO) measure and Bartlett’s test of sphericity (p<0.001) were used to assess the appropriateness of the sample size for factor analysis.

The excluded factors from the factor analysis were those that did not increase behaviour variance. Cronbach’s α coefficient values were used to assess the internal consistency of the Stretching Exercise Influencing Scale (SEIS). Intraclass correlation coefficient (ICC) was done with 30 computer users who completed the questionnaire twice at a 2-week interval. The acceptable value for ICC was considered 0.4 or above. Data analyses were undertaken using the Statistical Package for Social Sciences. The frequency/percentage and mean (SD) for analysing demographic variables were used.

Patient and public involvement

Patients and/or public were not involved in the designing and planning of the study.

Results

In all, 420 office employees including 113 men (26.9%) and 307 women (73.1%) participated in the study. Table 1 shows the demographic characteristics of the participants. The KMO measure was 0.914, which fell in the ‘very good’ category. Bartlett’s test of sphericity was meaningful (p<0.001) which indicates that the sample size was sufficient for EFA. Through EFA, from the primary 86 items, 9 items were not loaded on any factor and were removed. The initial analysis indicated an 11-factor structure with 77 items for the questionnaire with a total score between 77 and 293. All the remaining items were found to have significant item-to-total correlations (p<0.05). Table 2 shows the main factor analysis of the varimax rotation for the questionnaire. Table 3 shows all 11 factors and their reliability characteristics. All 11 factors had real commonalities (the subscales ranged between 0.73 and 0.89). Cronbach’s α coefficient for SEIS was 0.84 with a satisfactory result.

Test-retest of the scale at a 2-week interval was done on 30 computer users. All computer users complied with that because all were working and available in the office after 2 weeks. The results of ICC indicated appropriate and acceptable stability (ICC=0.78, 95% CI 0.70 to 0.88). The SEIS showed well constructed reliability and validity.

Discussion

This study developed and evaluated the psychometric properties of SEIS among a sample of Iranian computer users. The 11-factor structure of SEIS was consistent with the original constructs of HPM and other evidence-based constructs. This well-constructed 11-subscale instrument may be due to good items that were based on good literature review and good experience of researchers regarding not practising SE in workplaces in Iran. The large sample size (n=420) of this study may result in good response for the designed instrument.

The internal consistencies of SEIS’ subscales were also similar to those demonstrated by other studies.9 15 16 Furthermore, in this study, explanatory factor analysis showed that the factors of perceived barriers to action, perceived self-efficacy and commitment to plan of action had satisfactory loading and contributed to doing SE. These findings are in the line with that of another study which found that commitments to other preferences prevent individuals from doing exercises in the workplaces, while perceived adherence to plan caused home exercise motivation.17 Another study revealed that commitment to plan of action is a key concept of HPM that could influence behaviour.10 These evidences support the results of the present study with regards to the validity of SEIS. However, the current study relies on the fact that self-regulation, counterconditioning and stimulus control construct were satisfactorily loaded in the instrument which influences preferences. These findings are in line with other evidences that argue with the positive impacts of these factors on the construct of preferences.18

In SEIS, there was a positive relationship between perceived benefit and doing SE that is supported by the results from other studies.18–20 Moreover, in the present study, perceived barrier and self-efficacy levels were found to be effective for SE. This result is consistent with the confirmatory factor analysis of HPM in Robbins’ study in which social support structures, perceived barriers and self-efficacy were fit and significantly correlated with physical activity.21 It is well known that the perceived barriers to action could demotivate individuals' behaviour, so it is most important. Similar to the present study, a previous study stated that self-efficacy in physical activity could overcome external and internal barriers.22 Sharma, in his study, reported that physical activity interventions need to be built on promoting self-efficacy.23

Previous evidence reported the satisfactory validity and reliability for self-efficacy in the exercise scale among older adults.24 In our study, the instrument jointly accounted for 53.32% of the total variance for doing SE, which is well above the earlier studies assessing the model without the three constructs. Furthermore, it was determined that the structure of the instrument consisting of 11 factors and 77 questions explained desirable variance for doing SE. Zheng’s and Newman’s studies showed 57% and 71% of the variance in adherence to exercise, respectively, both of which are higher rates compared with our study.10 13 While our analysis suggested that the SE scale showed good reliability and strong internal consistency, Rivière’s study25 showed poor-to-good reliability, credibility and concurrent validity.

This study designed and validated an SEIS among Iranian office employees. According to the findings, satisfactory psychometric properties for the instrument were achieved. This achievement regarding good factor recovery may be due to adequate sample size (420 individuals) in this study, although the limitation of small sample size has been mentioned in other study.26

WMSDs of different employees were not specifically the same27. WMSDs are a multidisciplinary problem and biopsychosocial demographic characteristics may affect it.28 29Moreover, no analysis was done to realise the differences between the subgroups in terms of marital status and educational level. In spite of these differences, a questionnaire with a good recovery factor could be obtained because of the general similarities between the reasons for not doing SE at the work site.

The results of this study are not representative of the general population due to sampling from only one university and also because the majority of the participants was aged ≥41 years. However, despite these probable limitations, the designed scale could determine the factors that may have an impact on doing SE among the target group.

Conclusion

The designed scale in the present study could determine the factors which may have an impact on doing SE among a sample of Iranian computer users. Therefore, this study could be a foundation for further investigations for confirming this instrument as a culturally appropriate tool for assessing factors that may influence SE behaviour.