Participants characteristics
A total of 179 participants were recruited from EDs in private, teaching, and governmental hospitals in Amman to fill out the translated questionnaire. The mean age of the participants was 42.71 years (SD = 17.75, Range = 18–87), and more than half were male. The patients were recruited during various shifts and at different triage levels. Detailed information on participants’ characteristics is presented in Table 1.
Content validity assessment
The I-CVI ranged between 0.66 and 1. According to Polit et al. [19], items with an I-CVI of 0.78 or higher for three or more experts indicate good content validity. Accordingly, all items except item 12 demonstrated acceptable validity (I-CVI = 0.66). The S-CVI was 0.96. This suggests that the scale items were relevant to the scale and the scale was appropriate for the measurement purpose.
Pilot testing
All the pilot sample (15 patients) found the instrument easy to use, with clear and understandable instructions. However, four participants (26%) had to clarify question number 12, which was rephrased by the research team. Furthermore, the time required to complete the questionnaire was determined to be 8–10 min.
Construct validity assessment
Factor analysis
An initial statistical model (CFA) was tested to examine the three subscales in the original PCCC-R. This model did not provide an adequate fit with the hypothesized factor structure of the Arabic-PPPC-R, as indicated by several statistical measures, suggesting the model needs refinement. The chi-square value was significant: χ2 = 261.438, χ2/df = 2.16. The Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) values were 0.889 and 0.871, respectively, which are below a threshold of 0.90 [20]. The Root Mean Square Error of Approximation (RMSEA) was 0.089 with a 90% CI of 0.073 to 0.105, which is higher than a threshold of 0.08 [21].
Therefore, the researchers conducted an exploratory PAF analysis using Promax rotation to examine item loading. The KMO measure of the sampling adequacy was 0.93, and Bartlett’s test of sphericity was significant at p <.001, reflecting that the sample used to check the construct validity in the current study was adequate [17]. The PAF using Promax rotation initially identified three factors with eigenvalues greater than 1; examining the scree plot supported our decision to keep three factors (Fig. 2).
Scree plot of the items’ loading
The extracted factors represent three subscales that explain 66.3% of the variance (Table 2). The first factor, labeled “Context and Relationship,” comprises seven items, with an eigenvalue of 9.55, accounting for 53.06% of the variance. The factor loadings of the items ranged from 0.539 to 0.770. The second factor is the “Healthcare Process,” comprising five items; this factor had an eigenvalue of 1.33 and explained 7.08% of the variance. The factor loadings of the items ranged from 0.472 to 0.760. The third factor is the “Treatment plan,” comprising five items; this factor had an eigenvalue of 1.11 and explained 6.11% of the variance. The factor loadings of the items ranged from 0.550 to 0.651. Item 12 in the original questionnaire did not have sufficient clear loading on any of these factors and was eliminated from the final version of the questionnaire. Also, the low I-CVI supported the decision to eradicate item 12.
Then, the new model was tested using CFA. Initially, we examined the normality distribution, which revealed univariate normality (Critical ranges (Cr) between − 1.96 and 1.96); however, there was multivariate Kurtosis (Cr = 12.5). The proposed adapted model showed acceptable fit measures: χ2/df = 2.01. The CFI and TLI values were 0.94 and 0.93, respectively. The RMSEA was 0.07 with a 90% CI of 0.061 to 0.09, and the Standardized Root Mean Square Residual (SRMR) was 0.0421. The standardized estimates of the model’s items are presented in Fig. 3, showing relatively high loadings, ranging from 0.64 to 0.84, while the correlations among the factors ranged from 0.80 to 0.87.
The final Confirmatory Factorial Analysis for the Arabic Revised Patient Perception of Patient-Centeredness scale (displayed question numbers represent numbers in the original questionnaire)
Reliability testing
The reliability analysis of the Arabic-PPPC-R Questionnaire, based on a sample of 179 participants, demonstrated strong internal consistency. The Cronbach’s alpha coefficient for the Arabic-PPPC-R questionnaire was 0.949. Additionally, Cronbach’s alpha was calculated for each subscale: Context and Relationship (α = 0.889), Health Care Process (α = 0.890), and Treatment Plan (α = 0.906). The corrected item-total correlations for individual items ranged from 0.580 (PPPC 14) to 0.761 (PPPC 3), indicating good item discrimination. Cronbach’s alpha if an item was deleted remained consistently high, ranging from 0.945 to 0.948, suggesting that all items positively contributed to the overall reliability of the scale. These results support the internal consistency of the Arabic-PPPC-R, with no item significantly weakening the scale’s reliability.
link