Machine translation post-editing for cross-cultural adaptation of a patient safety culture instrument: Implications for nursing education.

AIM: To examine the feasibility of integrating machine translation post-editing (MTPE) into the cross-cultural adaptation of the Hospital Survey on Patient Safety Culture version 2.0 (HSOPSC 2.0) and to conduct a pilot psychometric evaluation of the resulting Traditional Chinese version (TC-HSOPSC 2.0).

DESIGN

A two-phase methodological validation study.

METHODS

From May to July 2025, HSOPSC 2.0 was translated and culturally adapted into Traditional Chinese using an MTPE workflow incorporating ChatGPT-4, Google Translate, BLEU-based screening, expert review, and iterative post-editing. Face validity was evaluated through cognitive interviews with 12 clinical nurses, followed by pilot psychometric testing with 32 clinical nurses. Reliability was assessed using Cronbach's alpha, ICC, and weighted kappa.

RESULTS

Initial back-translation screening indicated that 68.3% of translated items exceeded the predefined BLEU criteria. Following expert review and iterative revision, this proportion increased to 70.7% for Google Translate and 80.5% for ChatGPT-4. Face validity assessment identified five items requiring revision, and item-level face validity indices improved from 0.58-0.67 to 0.92-1.00 after cognitive interviewing and expert review. The final TC-HSOPSC 2.0 demonstrated excellent internal consistency (Cronbach's α = 0.93) and good test-retest reliability (ICC = 0.74), with item-level weighted kappa coefficients ranging from 0.21 to 0.74, indicating fair-to-substantial agreement.

CONCLUSIONS

The Nursing Education-Based Patient Safety Culture Assessment Tool was successfully translated and culturally adapted using an MTPE approach. The findings demonstrate that combining AI-assisted translation with expert review and cognitive interviewing can ensure translation quality and cultural relevance. This MTPE workflow provides a feasible and rigorous strategy for adapting nursing assessment instruments across diverse linguistic and cultural contexts.

IMPACT: The MTPE workflow may facilitate the efficient adaptation of nursing and healthcare measurement instruments, thereby supporting the implementation and evaluation of evidence-based educational initiatives across diverse linguistic and cultural contexts.