Development of a network for interactions and associations among biopsychosocial features of chronic low back pain

Abstract Background Understanding the associations among biopsychosocial factors is essential for improving research and treatment of chronic low back pain (CLBP). Here we characterized interrelations among biopsychosocial domains using network analysis and identified the most influential features in CLBP. Methods Data came from Quebec Low Back Pain Study, comprising 4,489 CLBP participants. We modeled relationships among baseline biopsychosocial features as networks, where nodes represent features and edges encode statistical or causal dependencies among them. Undirected network was inferred using distance correlation. Directed network was constructed using the Linear Non-Gaussian Acyclic Model, which estimates plausible causal directions. Influence maximization was performed using the Independent Cascade (IC) model to identify the most influential features in each network. Results In the undirected network, physical function and pain interference were the most central nodes, followed by depression. In the directed network, fear of movement, catastrophizing, and widespread pain emerged as key downstream hubs receiving multiple causal inputs, whereas pain interference, physical function, and depression acted as major upstream drivers exerting broad causal influence. IC diffusion simulations further identified pain interference and physical function as the most influential features in the undirected and directed networks, respectively. Conclusions Pain interference, physical function, and depression consistently emerged as key components of the CLBP biopsychosocial network. These features exert causal effects on fear of movement, catastrophizing, and widespread pain, with diffusion analyses confirming their roles as system-wide drivers. Interventions targeting functionality and pain interference, rather than pain intensity alone, may yield broader benefits across psychological and functional domains.