Participants and data collection
In this cross-sectional study, we recruited a convenient sample of 450 RA patients referring to a rheumatology clinic in Hamadan, Iran, to participate in the study. This nine-month study was conducted from May 2017 to February 2018. We estimated sample size based on 15 samples per item [27]. The primary version of scale comprised 30 items, and thus, a sample of 450 participants was considered as respondents. Inclusion criteria were (1) having at least four diagnostic criteria of RA, as suggested by American Rheumatology Association, (2) suffering from RA for more than 6 months, (3) being with at least 18 years of age and, (4) having no psychological or audio-visual problems. We recruited patients who fulfilled the inclusion criteria, consecutively, until the planned sample size was reached. Fourteen patients rejected to participate in our study (Response Rate = 96.8%). So, we included the data on 436 patients into analysis. We explained the purpose of study, as well as the patients’ rights as human subjects for the research to the participants. All those who accepted participation signed consent forms. In order to collect data, the first author conducted face-to-face private interviews with all participants in a private room at the clinic. The mean time to complete interviews was about 30–35 min.
Measures
Self-Care Behaviors Scale developed by Morowatisharifabad et al. [26] was considered as the basis to develop the SCBS questionnaire. This scale included 17 items, within which the participants were requested to state “the frequency of performing various self-care activities for their arthritis on a regular basis (once a month) during the previous 12 months” [26]. A five-point Liker-type scaling ranged from zero (not at all) to four (always) was considered as the response format. The theoretical range for the scale was from zero to 68, within which the higher scores represented higher levels of performance in self-care behaviors.
Arthritis Self-Efficacy Scale (ASES) [28], the Persian version [26], was used to assess convergent validity of the SCBS. This scale comprises nine function items, five coping with pain items, and six items related to other RA symptoms (e.g. fatigue, depression). Due to logistical limitations, we chose to include the pain and other symptoms scale scores in the questionnaire, only. Response format was based on a four-point Likert-type scale: zero = not at all, one = seldom, two = sometimes, and three = a lot. The total score was in a range from 0 to 33, in which the higher scores indicated the higher levels of perceived self-efficacy among the patients.
Demographic Data Form was a nine-item scale developed by the researchers to collect data on socio-demographic characteristics of the respondents. The items included age, gender, occupation, marital status, level of education, household monthly income, residency place, disease duration and the history of RA in the family.
Content validity
The initial 17-item scale was reviewed and assessed by an expert panel consisting two rheumatologists, a sports medicine specialist, an internal specialist with field experience in RA, a psychologist, two scholars in the area of health behavior and education and two nurses with field experience in RA. Based on the primary idea of the experts, the scale was not comprehensive in terms of assessing all domains of self-care behaviors. They finally recommended us to extend the scale through conducting a fast literature review on new relevant studies. So, we conducted a literature review. Based on the search results [22, 23, 29,30,31,32,33,34], 13 items were found to be added to the initial 17 items. Therefore, the first draft of the SCBS comprising 30 items was developed. In a second occasion, the draft was presented to the expert panel. During panel, the items were reviewed and assessed, orally, and evaluated in terms of appropriateness and relevance of items to RA patients, and response format, as well.
Content Validity Index (CVI) and Content Validity Ratio (CVR) were applied to validate the content of scale, quantitatively. Eleven specialists in the areas of health education and health behavior, rheumatology, sports medicine, psychology and nursing were requested to apprise the necessity of each item on the basis of a 3-point Likert-type scale (it is necessary, it is useful but not necessary, it is not necessary) (CVR). The values more than 0.99 (based on the Lawshe table) for each item was considered as necessary for the scale. The eleven experts were also asked to assess clarity, relevancy, and simplicity of the items, on the basis of a 4-point Likert-type scale. For each item, we considered the CVI value greater than 0.79 as appropriate and acceptable. So, the items with the score less than 0.79 were deleted from the scale.
We interviewed the panel of experts face-to-face to assess the items’ level of difficulty. We asked them to report the level of importance for each item, using a 5-point Likert-type scale (not important at all, a little important, moderately important, important, and absolutely important). Then, we calculated the impact score of the items through multiplying the frequency of an item by its mean importance [impact score = frequency (%) × importance]. Eventually, the items with impact score ≥ 1.5 were considered for the next stage.
The final draft was, then, pilot tested among a sample of 41 RA patients. In the pilot study, we aimed to assess the utility of scale, to identify the benefits and problems associated to the design, and to estimate the internal consistency of the scale, using Cronbach’s alpha coefficient. We did not include the pilot sample in the final sample.
Translation into English
We translated the final version of SCBS into English, with the hope to be applied in future studies within different communities. As the SCBS was originally developed in Persian, we asked a native Persian speaker with mastery of the English language to translate it into English. In order to preserve the denotation and connotation of the items, we then back-translated [35] the scale into Persian by a native English speaker with mastery of the Persian language. The latter translator had not seen the original Persian version of the scale. Next, we compared the back-translated copy to the original Persian scale to recognize incongruities.
Statistics
We used the statistical Package for Social Sciences (SPSS) v. 22 for the purpose of data entry, manipulation and analysis. No item was found with missing data. We used measures of central tendency and variability to summarize and organize the data. We then performed Pearson’s Correlation Coefficient, EFA, CFA, and Internal Consistency Reliability tests. The level of significance was considered 0.05, a priori.
Construct validity
Applying principal component factor analysis with varimax rotation, we performed Exploratory Factor Analysis (EFA) to assess construct validity and factor structure of the scale. We also used Confirmatory Factor Analysis (CFA) with the robust maximum likelihood to estimate model parameters.
In order to determine the factor structure of the scale, an EFA was conducted based on a randomized split of the data in the sample. We randomly selected a sample of 200 participants using the randomization function on SPSS v. 22. During EFA, we considered the factor loadings equal or greater than 0.3 to be appropriate, and the eigenvalues above 1 as an assignment for the number of factors. We then used the Kaiser-Meyer-Olkin (KMO) and Bartlett’s Test of sphericity to obtain the appropriateness of sample.
Thereafter, we performed a CFA on the remaining 236 participants of the larger overall sample to identify whether the factor structure required modification. The Analysis of Moment Structures (AMOS), version 10.0 was applied to conduct the CFA. In the CFA process, the absolute fit of the model to data was evaluated using the χ2 statistic, the comparative fit index (CFI), the Tucker–Lewis Index (TLI), and the root mean square error of approximation (RMSEA) tests. We considered the model to be acceptable if χ2 was between 1 and 5, CFI was more than 0.8, TLI was greater than 0.9, RMSEA was < 0.05 good fit or between 0.05 and 0.08 adequate fit.
Reliability
We used Cronbach’s alpha test to investigate internal consistency of the scale. The Cronbach’s alpha coefficient of 0.7 or above was considered to be acceptable. We also applied Intra class Correlation Coefficients (ICC) to calculate the test-retest reliability coefficient (ICC ≥ 0.70 was considered satisfactory).
Convergent validity
We applied Pearson’s correlation coefficient test to assess the nature of associations between the SCBS factors, and to evaluate the associations between the factors and the domains of ASES.
Ethical considerations
Ethics committee in Tabriz University of Medical Sciences approved the study (number 40773, 16.11.2017). We obtained informed consent form from all respondents, and all signed consent forms. We also explained the patients about the purpose of study, and assured then on the confidentiality of their data.