Study design and sample
In recent years, the increase in the number of multicultural studies has urged the need to adapt scales to be used in other languages. Hence, depending on different cultures, the scales should be culturally modified and adapted. Regarding this point, cross-cultural adaptation must be applied as the study design. This consists of translation, adaptation, calculation of validity, and reliability. The present study was also a cross-cultural adaptation conducted in Shahid Chamran and Shadpour health complexes affiliated to Tabriz University of Medical Sciences in Tabriz, Iran 2015. Health complexes were launched after implementation of the health system reform plan in Iran, which assumed to provide integrated care in a defined area under district health center policies and regulations. Each health complex consists of 4–5 health centers that cover 10,000 to 30,000 people. Health centers generally provide preventive services and, if necessary, refer individuals to more specialized services to the health complex clinics.
Sampling and sample size calculation
Sample framework for selecting people was based on the registered records health. As the samples of this study, 220 older people who attend these complexes were recruited. The sample size of 220 people was calculated using G-Power software. Sample allocation to each complex was done in the same way. An a priori G-Power analysis was performed. Thus, considering independent samples t-test (two tailed) with 99% power and alpha = 0.01 significance level, we estimated an overall sample size of 220 older people for detecting differences between two independent means with effect sizes (d = 0.66). The participants were selected using a systematic random sampling method according to the following inclusion criteria: age over 60 years; ability to communicate; having a stable condition; and willingness to take part in this study. People, who were not permanent residents of Shahid Chamran and Zafaraniyeh or who suffered from acute disease, were excluded from this study.
Instrument
The SASE, as a self-reporting instrument, measures self-care ability among older people [25]. In fact, three versions were available in Chinese, Italian and Swedish. This instrument consists of two parts: the first part, the demographic characteristics of the elderly, and the second part consists of 17-item Likert scale that highlights areas of particular importance to self-care for the elderly, i.e. daily life activities, welfare, power, desires, determination, loneliness and wear clothes. The accountability scale of each item is between 1 “totally disagree” to 5 “totally agree”. A higher overall score indicates a higher self-care ability. The construct of self-care ability is considered as a necessary condition for self-care, and when this ability is exercised, self-care actions are achieved. The internal structure of SASE includes “ability for care of repertoire”, “ability for care of goal”, and “ability for care of well-being” [23, 24] .
Translation of the instrument
The permission to translate the English version of the SASE was obtained from the main author (Söderhamn). Then, the forward-backward translation and cultural adaptation was performed at several steps. A professional translator translated the instrument into Persian and another professional translator translated the Iranian version back into English. Both translators performed the translation process separately. After the completion of translation, an expert panel consisting of two elderly specialists and three elderly nursing compared the original English version with back translation and following cultural and linguistic adaptation, pre-final Iranian version of the SASE was prepared. Then, this version was examined among ten older people and modifications were done based on their feedback. At last, it was developed and used in this study for further psychometric assessment.
Statistical analysis
Descriptive statistics were used to describe the characteristics of the sample and several statistical analyses were applied to assess the psychometric properties of the Iranian version of SASE.
Content validity
The content validity was assessed through modified KAPPA by means of qualitative and quantitative approaches [27]. In this regard, the Iranian version of the SASE was sent to 10 specialists with sufficient clinical experience and theoretical knowledge about nursing and self-care. The fields of the experts were health education and promotion, geriatric health, community health, psychology, nursing, and epidemiology. In the qualitative approach, experts were requested to provide their own ideas for improving its quality. And in the quantitative approach, in order to calculate the CVI and modified KAPPA coefficient basis on relevance and clarity, each item was ranked on a four-point Likert scale based on views of the experts [27, 28].
Reliability
The reliability was assessed using internal consistency and the test–retest reliability method. The internal consistency was measured by the coefficient Cronbach’s alpha, which varies from 0 to 1, and values equal to or > 0.70 for a scale show a satisfactory internal consistency [29]. In order to estimate the test-retest reliability, the intra-class correlation coefficients (ICCs) during a two-week interval were calculated among 30 older people for total scale and each subscale. The following categories were chosen for the interpretation of agreement levels: 00–0.2 as small, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as remarkable and 0.81–1 as approximately perfect [30].
Exploratory factor analysis
Prior to EFA, Kaiser-Meyer-Olkin (KMO) was calculated to ensure that the sample size was adequate. KMO shows values between 0 and 0.49 as unacceptable, 0.5–0.7 as mediate; 0.7–0.8 as good, 0.8–0.9 as great, and > 0.9 as excellent [31]. EFA was performed using the principal component analysis with varimax solution [32]. Since we could not find a clear relationship between the items in the questions and the construct to be measured, we needed an exploratory factor analysis. In other words, we were not sure which items measured which structures. Therefore, because of this suspicion, we used exploratory factor analysis [33]. To select the factors, an eigenvalue greater than 1 and a factor loading equal to or greater than 0.4 were used. All the statistical analyses were performed using SPSS 21 statistical software package.