Cross-cultural adaptation and validation of the Hospital Survey on Patient Safety Culture 2.0 – Brazilian version

Abstract Background Patient safety culture concerns the values, beliefs and standards shared by an organisation’s health staff and other personnel which influence their care provision actions and conduct. Several countries have made a priority of strengthening patient safety culture to improve the quality and safety of health care. In this direction, measuring the patient safety culture through validated instruments is a strategy applied worldwide. The purpose of this study was to adapt transculturally and validate the HSOPSC 2.0 to Brazilian Portuguese and the hospital context in Brazil. Methods Of the various validated scales for measuring safety culture, the instrument most used internationally is the Hospital Survey on Patient Safety Culture (HSOPSC) developed by the United States Agency for Healthcare Research and Quality in 2004 and revised in 2019, when version 2.0 was released. Adaptation was conducted on a universalist approach and the adapted instrument was then applied to a sample of 2,702 respondents (56% response rate) comprising staff of a large general hospital in the city of São Paulo. Construct validity was investigated by Exploratory Structural Equation Modelling-within-Confirmatory Factor Analysis (ESEM-within-CFA) and reliability was measured in each dimension by means of Cronbach alpha coefficients. Results ESEM fit indexes showed good data fit with the proposed model: χ2 = 634.425 df = 221 χ2/df ratio = 2.9 p-value < 0.0000; RMSEA = 0.045 (90% C.I. = 0.041—0.050) and probability RMSEA < = .05 = 0.963; CFI = 0.986; TLI = 0.968. However, ten items had loads lower than 0.4. Cronbach alpha values were 0.6 or more for all dimensions, except Handoffs and information exchange ( $$\alpha$$ α = 0.50) and Staffing and work pace ( $$\alpha$$ α = 0.41). Conclusion The psychometric properties of the Brazilian version were found to be satisfactory, demonstrating good internal consistency and construct validity as expressed by estimates of reliability and indexes of model fit. However, given factor loadings smaller than 0.4 observed in ten items and considering that the scale translated and adapted to Portuguese was tested on a single sample during the Covid-19 pandemic, the authors recognize the need for it to be tested on other samples in Brazil to investigate its validity