The research team, led by Xuejie Yu and Wan Zhao of the Department of Otolaryngology-Head and Neck Surgery at the First Affiliated Hospital of the University of Science and Technology of China in Hefei, employed a suite of network analysis methods that have gained traction in psychological science over the past decade. Rather than treating symptoms as isolated variables to be correlated one by one, network analysis conceptualizes them as nodes in a web of mutually reinforcing relationships. The resulting structure can reveal which nodes sit at the center of the web, which ones connect otherwise separate domains, and which ones, if pushed in one direction, might drag the rest of the network along with them.

To build the network, the researchers assessed head and neck symptoms using the MD Anderson Symptom Inventory-Head and Neck Module, a validated instrument that captures the specific symptom burden of this patient population, including pain, swallowing difficulty, voice problems, dry mouth, and throat sores. Fear of cancer recurrence and related concerns were measured with the Fear of Progression Questionnaire-Short Form, which probes worries about disease progression, treatment, family impact, and the anxiety surrounding medical check-ups. The team then estimated the network structure using EBICglasso, an extended Bayesian information criterion graphical least absolute shrinkage and selection operator method that identifies the most important connections while filtering out spurious ones. The final network contained 239 edges, or connections, among the 34 nodes spanning both symptom and fear domains.

Three nodes emerged with the highest expected influence, a centrality metric that captures how strongly a node is connected to its neighbors and how much it might affect them if it changes: check-up anxiety, major treatment worry, and throat sores. This finding is notable because it places psychological concerns and physical symptoms side by side at the top of the influence hierarchy. Check-up anxiety, sometimes called scanxiety in the oncology literature, reflects the dread patients experience around follow-up appointments and imaging, moments when the possibility of recurrence becomes most concrete. That this node ranks among the most influential in the entire network suggests it may be a linchpin connecting the emotional and somatic dimensions of postoperative life.

The researchers also computed bridge expected influence, a measure designed to identify nodes that serve as conduits between distinct communities within a network. In this case, the bridge analysis pointed to disease progression worry, worry about family future, and check-up anxiety as the leading bridge nodes. These are the concerns that link the fear domain to the symptom domain, potentially allowing distress in one area to spill over into the other. If a patient’s worry about the cancer progressing intensifies, the bridge analysis suggests this could reverberate through the network and amplify physical symptom experiences, and vice versa. The identification of these bridge nodes has direct clinical implications: they represent potential leverage points where a targeted intervention might disrupt the cross-domain transmission of distress.

To move beyond descriptive network statistics, the team employed Bayesian network modeling, a probabilistic approach that estimates directional relationships between nodes rather than assuming all connections are symmetric. The Bayesian network identified seven cross-domain arcs, meaning seven directed connections that specifically linked head and neck symptoms to fear of cancer recurrence concerns. This directional structure is critical for intervention modeling because it suggests pathways along which change might propagate. The researchers then used Ising simulation, a computational technique borrowed from statistical physics where interacting units influence each other’s states, to model what would happen if specific nodes were aggravated or relieved across the entire 34-node network.

The simulation results are perhaps the most striking contribution of the study. When all 34 nodes were simulated for aggravation, swallowing difficulty yielded the largest point estimate, with a simulated increase of 19.28 percent and a 95 percent confidence interval ranging from 3.42 to 37.89 percent. In other words, the model suggests that if swallowing difficulty worsens, the overall network burden, encompassing both symptoms and fears, could increase substantially. Conversely, when nodes were simulated for relief, choking cough produced the largest point estimate, a decrease of 17.51 percent with a confidence interval from negative 29.81 to negative 4.26 percent. These two findings converge on a common theme: the mechanical act of moving food and liquid safely from mouth to stomach, which is compromised in many laryngeal cancer patients after surgery, appears to be a central node whose state influences the broader psychological and somatic network.

The clinical significance of swallowing difficulty in head and neck oncology is well documented in prior research. Studies have shown that swallowing function after laryngeal cancer treatment is a major determinant of quality of life, nutrition, and social functioning. Patients who cannot swallow comfortably often withdraw from social eating situations, which can compound feelings of isolation and anxiety. The network analysis framework adds a new dimension to this understanding by suggesting that swallowing difficulty is not merely one symptom among many but a potential driver of the entire symptom-fear constellation. Prophylactic swallowing interventions, which have been reviewed extensively in the nursing literature, may therefore have benefits that extend beyond the mechanical improvement of swallowing itself, potentially dampening the fear of recurrence and related psychological distress.

Choking cough, the node with the largest simulated relief effect, is equally compelling from a clinical standpoint. After total laryngectomy or other laryngeal surgeries, patients lose the normal protective mechanisms of the upper airway, making aspiration and coughing during swallowing common and distressing events. The heat and moisture exchanger, a device studied in randomized controlled trials for pulmonary protection after laryngectomy, represents one approach to managing airway complications. The simulation finding suggests that interventions targeting choking cough, whether through swallowing therapy, dietary modification, or airway protection strategies, might produce outsized benefits across the network, reducing not only the physical symptom itself but also the anxiety and fear that it triggers.

The authors are careful to frame these findings as candidate targets rather than proven intervention points. The study is cross-sectional, capturing a single moment in time rather than tracking changes longitudinally, which means the simulated effects represent model-based hypotheses rather than demonstrated causal relationships. The confidence intervals around the point estimates are wide, reflecting uncertainty inherent in the simulation approach. Nevertheless, the convergence of multiple analytical methods, including expected influence, bridge expected influence, Bayesian network modeling, and Ising simulation, on a coherent set of findings lends weight to the conclusions. The identification of disease progression worry, worry about family future, and check-up anxiety as bridge nodes, combined with the simulation results pointing to swallowing difficulty and choking cough, offers clinicians a structured framework for postoperative assessment that considers head and neck symptoms and fear of cancer recurrence together rather than in isolation. As supportive care in oncology increasingly embraces personalized symptom management, this network-based approach may help identify which patients need which interventions most, and where those interventions are likely to have the greatest ripple effect across the complex web of postoperative experience.

Subject of Research: Network analysis of head and neck symptoms and fear of cancer recurrence in postoperative laryngeal cancer patients: simulation-based intervention modeling

Article Title: Network analysis of head and neck symptoms and fear of cancer recurrence in postoperative laryngeal cancer patients: simulation-based intervention modeling

Article References: Yu, X., Ding, C., Chen, W., Yang, Y., Li, X., Zhang, Y., & Zhao, W. (2026). Network analysis of head and neck symptoms and fear of cancer recurrence in postoperative laryngeal cancer patients: simulation-based intervention modeling. Supportive Care in Cancer, 34(10), Article 930. https://doi.org/10.1007/s00520-026-11172-1

Image Credits: AI Generated

DOI: 10.1007/s00520-026-11172-1

Keywords: Network, analysis, head, neck, symptoms, fear, cancer, recurrence, postoperative, laryngeal, patients, simulation-based