How the psychological safety of healthcare teams relates to team performance, specifically patient outcomes in an acute private hospital in Mpumalanga, South Africa

Abstract

The primary purpose of this study was to investigate the influence of psychological safety on patient safety performance within an acute private hospital in Mpumalanga, South Africa. In the context of a resource-constrained and hierarchical healthcare environment, the study sought to identify the specific mechanisms that translate interpersonal safety into clinical excellence. A quantitative, cross-sectional design was employed, utilising a structured survey administered to 75 healthcare professionals, including nurses, doctors, and allied staff. Partial Least Squares Structural Equation Modelling (PLS-SEM) was used to test the hypothesised relationships. The findings revealed that inclusive leadership is a robust predictor of psychological safety ($\beta=0.750$), confirming that safety is a ""top-down"" permission structure in this context. Contrary to initial assumptions, psychological safety did not have a significant direct effect on patient safety performance. Instead, the relationship was fully mediated by team learning behaviours. This indicates that psychological safety improves patient outcomes only when it activates learning processes such as error reporting and reflection. While traditional literature often positions PS as a direct predictor of performance, this study reveals more complex realities within the South African context. Findings indicate that PS alone does not improve patient outcomes; rather, it acts as a catalyst for team learning behaviours. Based on these findings, it is recommended that hospital management shift focus from generic safety culture campaigns to specific inclusive leadership training and the institutionalisation of routine shift debriefs. Future research should employ longitudinal designs to measure the long-term impact of these interventions on objective clinical outcomes. The study is limited by its cross-sectional nature and sample size, though statistical power remained high.