Network analysis of anxiety, depression, stress symptoms and psychological resilience in shift nurses.

objectiveTo explore the network structure of anxiety, depression, and stress symptoms in shift nurses and their relationship with psychological resilience using network analysis and exploratory probabilistic modeling methods, providing a foundation for developing targeted prevention and intervention strategies.

methodsA cross-sectional survey was conducted among 753 shift nurses from three tertiary hospitals in eastern China. DASS-21 was used to assess anxiety, depression, and stress symptoms, and CD-RISC-10 was used to assess psychological resilience. Graphical LASSO (gLASSO) was used to construct symptom networks, expected influence (EI) and bridge expected influence (BEI) were calculated to identify core symptoms and bridge symptoms, Bayesian network analysis was used to explore potential conditional dependencies between symptoms, and flow network analysis was used to explore the association between psychological resilience and symptom networks.

resultsThe anxiety, depression, and stress symptom network of shift nurses included 21 nodes and 100 non-zero edges (i.e., edges retained after LASSO regularization). Network centrality analysis identified “Scared without reason,” “inability to experience pleasure,” “difficult to relax,” and “energy expenditure” as core symptoms, and “inability to experience pleasure,” “lack of initiative,” and “difficult to relax” as key bridge symptoms. Bayesian network analysis showed that “no expectations,” “mental effort expenditure,” “difficult to relax,” and “intolerant of interruptions” occupied conditionally prior positions in the directed acyclic graph, suggesting potential directional dependencies in the exploratory model. Flow network analysis showed that psychological resilience was significantly negatively associated with 12 symptoms.

conclusionNegative emotional symptoms of shift nurses form a complex interactive network. Core symptoms and bridge symptoms show strong statistical connectivity within the estimated network. Bayesian network modeling suggests potential conditional dependencies that warrant further investigation through longitudinal designs. Psychological resilience shows negative associations with multiple core and bridge symptoms, suggesting potential protective effects. These findings are hypothesis-generating and require validation through prospective and intervention studies. (The First Affiliated Hospital of Nanjing Medical University Approval No. 2025-SR-1217)