Instrument design
The researcher performed a substantial literature review to identify the different dimensions used in assessing the use of telemedicine in different countries (Barsom et al., 2020; Chau & Hu, 2002; Dhukaram et al., 2011; Di Cerbo et al., 2015; Peng et al., 2020). The instrument items were adapted from these findings. The survey is attached in the supplementary file.
The questionnaire was generated in two main steps. The first selection of questions was related to assessing intention to use a consultation. This step was performed by reviewing the literature; as a result, the majority of the literature followed the Technology Acceptance Model (TAM) in generating the questionnaire with specific factors: perceived usefulness, which is explained by motivation, attitudes toward use, trust, and perceived ease of use (Chau & Hu, 2002; Gurupur et al., 2017; Vidal-Alaball et al., 2020) (Figure 1). Some elements were added, such as social factors, as important elements to be addressed when assessing patients’ behavioural intention to use consultations (Dhukaram et al., 2011; Vidal-Alaball et al., 2020). The study model was built as presented in Figure 1.
The second step was to select the scale, and based on the literature, a common type of scale used is the 5-point Likert scale: 1 strongly disagree, 2 disagree, 3 neutral, 4 agree and 5 strongly agree (Bakken et al., 2006; Bergmo et al., 2005; Gurupur et al., 2017; Keely et al., 2013)
Face and content validity
A face and content analysis was performed to match the study objectives. The reviewers were academic experts in the field of healthcare quality. The questions underwent some grammatical modifications and some rephrasing, which occurred for two study items: “I would characterize e-consultations treated providers as honest” was rephrased to “The treating physicians in the e-consultation services are honest” and “My GP does not offer e-consultation” to “The physician/hospital does not offer e-consultation”.
The questions were translated using backward and forward translation in English and Arabic using an accredited agency for translations. Then, the questions were reviewed again by the academic reviewer, and a sample of patients was collected to ensure that the questions were clear. The questions were approved and ready to be used.
The questions were distributed to all of the patients through social media, such as Twitter and WhatsApp, using Question Pro software.
Steps of content validity using factor analysis
- Exploratory factor analysis (EFA)., i.e., general loading to decide which item is related to each factor using principal component analysis and varimax rotation
- Confirmatory factor analysis (CFA): to assess the model stability of the suggested model from step 1. The confirmatory analysis was conducted following structural equational modelling (SEM) using AMOS. The SEM was extracted using maximum likelihood (ML) robust extraction methods.
Based on the following indices, the model fit criteria were assessed/
- Normed (X2/df) recommended to be less than 3 to be considered as a good fit
- RMSEA: the root mean-square error of approximation value recommended to be a value between 0.05 and 0.08, suggesting reasonable fit; a value >1.0 suggests poor fit, and values between up to 0.08 (Browne & Cudeck, 1992; Whittaker, 2016) or .10 (Whittaker, 2016) are considered acceptable.
- The comparative fit index (CFI) and Tucker-Leis index (TLI) show the research model with the baseline model, and a value >0.9 indicates reasonably good fit. (Bentler, 1990)
- Reliability: The determination of reliability was undertaken using Cronbach’s alpha coefficient to test the internal consistency of the responses for each dimension and the entire instrument.
Statistical analysis
The sample size was estimated to be 384 based on the following formula:
![](https://myfiles.space/user_files/58892_7798ecd9a40b82f9/58892_custom_files/img1623679990.png)
where p= for the population is estimated to be more than 1 million participants, e= margin of error is 0.5, and the t value is 1.96 (Taherdoost, 2017). For factor analysis studies, it is acceptable to have a sample of more than 50 participants, which can be estimated by the number of study items/independent variables (22) times (10 participants), so a total of 220 participants will be valid to run factor analysis statistics (Kotrlik & Higgins, 2001). In this study, 244 participants were included. Data are reported with 95% confidence intervals. All analyses were performed using SPSS IBM software, version 27, and AMOS software for confirmatory analysis.