Anxiety symptoms as the central node: a network analysis of mental health sequelae in a post-COVID-19 national sample

Background The COVID-19 pandemic has precipitated a global mental health crisis, with the long-term effects of Post-COVID-19 syndrome presenting a complex interplay of physical and psychological symptoms. However, the dynamic network interrelationships between social support, perceived stress, and general mental health and how these networks vary across demographic groups remain inadequately explored, hindering the development of targeted interventions. Methods To address this gap, we constructed an interactive network model using national survey data ( N = 7,997). A Gaussian Graphical Model (GGM) was employed to estimate partial correlations among variables, and network comparison tests were conducted to examine structural differences across gender, income, residence, and marital status. Results Anxiety and insomnia symptoms (GHA, from the General Health Questionnaire) exhibited the highest strength centrality in the network, indicating their position as the most statistically interconnected node. A strong negative correlation was identified between social support (SSS) and feelings of helplessness (PLC, from the Perceived Stress Scale), underscoring the potential protective role of social support. Notably, network invariance tests revealed significant structural variations across demographics with meaningful effect sizes. Women showed stronger stress-depression connections, whereas men exhibited a stronger negative link between social support and helplessness. Low-income groups demonstrated tighter anxiety-depression connectivity; distinct network topologies were observed between urban and rural residents, and marital status differentially influenced the prominence of tension-related versus helplessness-support connections. Conclusion Mental health issues in the post-pandemic era exhibit distinct networked characteristics, with social support serving as a key buffer against stress. The substantial variation in symptom networks across demographic lines underscores the necessity of developing tailored, precision interventions for specific populations.