Objective To explore the network structure of factors associated with nurses’ willingness to participate in online Traditional Chinese Medicine (TCM) nursing services using network analysis, and to examine dose–response relationships between key variables and behavioral intention using restricted cubic spline analysis. Methods A cross-sectional survey was conducted in May 2024, recruiting 346 nurses from Beijing through convenience sampling. A self-developed 5-point Likert scale questionnaire collected information on knowledge level (10 items), risk perception (4 items, reverse-scored), technical willingness (7 items), perceived usefulness (3 items), self-efficacy (5 items), subjective norms (3 items), and behavioral intention (3 items). Network structure was estimated using EBIC-LASSO regularization. Four centrality indices (strength, betweenness, closeness, and expected influence) were calculated, with Bootstrap stability testing ( n = 1,000). Restricted cubic splines with 4 knots explored nonlinear relationships between continuous variables and intention. Results Among 346 nurses, 92.8% were female, 75.1% held bachelor’s degrees, and 95.1% had received TCM nursing training. Mean scale scores were: knowledge level 2.51 ± 1.08, risk perception (reverse-scored) 3.35 ± 1.12, technical willingness 2.04 ± 0.89, perceived usefulness 2.02 ± 0.88, self-efficacy 2.20 ± 0.92, subjective norms 2.19 ± 0.92, and behavioral intention 2.23 ± 0.96. Network analysis revealed self-efficacy as the core node with the highest strength centrality (1.371) and expected influence (1.298). The strongest associations emerged between self-efficacy and subjective norms (edge weight = 0.439) and between self-efficacy and behavioral intention (edge weight = 0.433), forming a core cluster. Risk perception was marginalized in the network (strength centrality = −1.757). Bootstrap testing showed strength centrality CS coefficient of 0.751 (0.5). RCS analysis demonstrated linear positive correlations between knowledge level, self-efficacy, perceived usefulness and intention (nonlinearity p 0.05), while risk perception exhibited a significant inverted U-shaped nonlinear relationship with intention ( p = 0.023). Conclusion Self-efficacy emerged as the most central variable in the estimated network and may represent an important focus for future research. The inverted U-shaped association between risk perception and intention suggests that both excessively high and excessively low levels of risk concern may be unfavorable for willingness formation.
Zhang et al. (Wed,) studied this question.