Ensuring Taiwan patient safety culture survey is free from differential item functioning

Abstract

Background Differential item functioning (DIF) means the interference of some demographic characteristic or grouping of the tight relationship between trait levels. DIF threatens precise interpretations of survey results and makes them unreliable. The aim of this study was to examine whether the succinct version of Taiwan Patient Safety Culture (TPSC-S) survey is free from DIF and to mitigate its impact if needed. Methods The TPSC-S survey results of 2964 respondents in a public hospital in Taiwan were analyzed. The existence, type, and effect size of DIF were examined for each TPSC-S item using a proportional-odds logistic regression method between characteristic groups, including gender, work experience, job types, management roles, employment status, and safety reporting experiences. Results The study results revealed that several items of TPSC-S showed statistically significant DIF between characteristic groups. Nevertheless, the magnitude of these DIF was small, and their influence to TPSC-S survey was not significant. The domain-level DIF impact was completely insignificant for all characteristic groups. Conclusion This study revealed that the 24-item TPSC-S survey was free from DIF in six characteristic groups. The difference in survey scores between groups stems from the real difference that hospital safety managers want to measure.