Psychometric Evaluation of the Korean Version of Hospital Survey on Patient Safety Culture

Abstract

Objectives This study evaluated the psychometric properties of the Korean-language version of the Hospital Survey on Patient Safety Culture (HSOPSC) among Korean nurses. Methods We analyzed secondary data from 801 direct care nurses working at a tertiary, private, nonprofit, teaching hospital in South Korea. Descriptive statistics, internal consistency coefficients, and intercorrelations were calculated. The latent factor structure of the HSOPSC was examined using exploratory structural equation modeling techniques, which account for the noncontinuous nature of ordinal data. Results Although a majority of subscales showed acceptable to good internal consistency, 4 dimensions (staffing, overall perceptions of patient safety, organizational learning-continuous improvement, and nonpunitive response to errors) had reliability levels less than 0.6. The HSOPSC items loaded somewhat diffusely on 3 subscales: staffing, teamwork across units, and organizational learning-continuous improvement. Correlations between the 12 HSOPSC factors indicated discriminant validity. Convergent validity was supported by correlations between the 12 subscales and a single-item outcome variable, namely, patient safety grade. Several items did not load well on their respective subscales, but most items fit the underlying theoretical model implied by the HSOPSC, resulting in an acceptable model fit (confirmatory fit index = 0.985, root mean square error of approximation = 0.034, weighted root mean square residual = 0.54). Conclusions Despite the acceptable model fit of the Korean version of HSOPSC, the psychometric properties of this instrument require further investigation to ensure it is an effective tool to measure patient safety culture and identify areas for improvement in the Korean health care system.