General patient characteristics
The study population included 88 males (28.7%) and 218 females (71.3%), with a mean age of 45.91 ± 16.60 years. Among the participants, 202 individuals (66.0%) were married, 92 (30.1%) were unmarried, and 12 (3.9%) were widowed or divorced. Additionally, 260 individuals (84.9%) were employed and 46 (15.1%) were unemployed. The sites of surgery included the abdomen (n = 126, 41.2%), breast or ovary (n = 60, 19.6%), knee (n = 32, 10.5%), hip (n = 32, 10.5%), lower back (n = 36, 11.7%), and shoulder (n = 20, 6.5%).
Validity analysis
Content validity
Almost all items (26/27) showed an I-CVI ranging between 0.80 and 1.00 and S-CVI/Ave of 0.90, indicating adequate content validity. The item “I looked around the operating theater” had an I-CVI of 0.80 and S-CVI/Ave of ≤ 0.90. However, it was determined that this was a coping behavior in patients during anesthesia surrendering, and consequently, we decided to retain all 27 items.
Item analysis
We analyzed the skewness and kurtosis to check the normality of the data used in this study. The skewness of all measured variables was between 0.01–1.15 (absolute value) and did not exceed 3; kurtosis did not exceed 10 and had absolute values between 0.04–1.41. The data showed normal distribution. However, the multivariate normality test showed that the multivariate kurtosis was 43.5 and the critical ratio was 14.2 (Table 1). It is not practical to confirm the combined frequency distributions for all variables, and few data satisfy the assumption of multivariate normality. Therefore, if there is no major problem with a univariate normality test, it can be assumed that multivariate normality is satisfied [27]. In summary, the measured variables in this study satisfied univariate normality, and thus, the validity of the ASI was satisfactory.
Construct validity
This study determined the number of factors through classification of more than one eigenvalue, parallel analysis, and Velicer's minimum average partial analysis. We found four factors using eigenvalues, two factors using parallel analysis, and two factors using Velicer's minimum average partial analysis. Based on the number of factors used for the original ASI, we finalized four factors for the factor analysis. EFA was performed for all 27 items, applying the principle component analysis and varimax rotation. The KMO value for determining the suitability of samples for factor analysis was 0.85, which was higher than the cutoff value of 0.80. The approximate χ2 value for the Bartlett’s test of sphericity was 2942.88 (df = 231, p < 0.001), which indicated that the samples were suitable for factor analysis. Factor analysis of all 27 items showed four factors with an initial eigenvalue ≥ 1.0. Among all times, the item “I looked around the operating theater” was excluded due to a commonality of 0.10, whereas all other items showed commonality ≥ 0.50, satisfying the factor loading criterion of ≥ 0.40. Based on the EFA results, 26 out of the 27 initial items were retained (Table 2).
Confirmatory factor analysis
CFA was performed for a total of 26 items under the four factors identified by preliminary construct validity testing. Goodness-of-fit assessment showed χ2 = 1022.33 (df = 293, p < 0.001), WRMR = 1.05, TLI = 0.92, CFI = 0.93, and RMSEA = 0.056 (95% CI = 0.053–0.059). These results indicated that the goodness-of-fit was satisfactory according to our criteria. The final model was used for convergent validity and discriminant validity testing. The standardized regression coefficients of all items were statistically significant, ranging between 0.50 and 0.97, and the ranges of CR and AVE were 0.85–0.92 and 0.50–0.61, respectively (Table 3). Moreover, when the inter-factor correlation coefficients were examined to test the discriminant validity of factor construct, they were ≤ 0.80, and the square inter-factor correlation coefficients were lower than the AVE values, thereby confirming the discriminant validity of the factors (Table 3).
Criterion validity
For criterion validity, the correlation between the Korean ASI and STAI-KYZ was verified; the results showed a negative correlation (r = -0.50 to -0.62, p < 0.001; Table 4).
Reliability analysis
The values of Cronbach’s alpha ranged between 0.71 and 0.88 for the dimensions used in the instrument, as shown by the reliability analysis for the internal consistency of the Korean ASI (Table 4).