Symptom Structure, Heterogeneity, and Network Features on Postoperative Day 3 After Oesophagectomy: A Multisite,...

The acute inpatient symptom state following oesophagectomy is multidimensional and heterogeneous. However, the structure, between-patient heterogeneity, and network features of symptoms remain incompletely characterised.

We analysed 1047 postoperative patients with oesophageal cancer from four hospitals in Anhui Province, China, in the "Be Resilient to Oesophageal Cancer" program. We applied principal component analysis to summarise symptom dimensions and latent profile analysis to derive patient groups of symptom burden. Symptom networks were estimated and centrality computed to identify candidate targets. We then performed in silico node-perturbation simulations to estimate potential system-level effects of alleviating versus aggravating individual symptoms.

Four symptom dimensions emerged from the principal component analysis: somatic discomfort and fatigue, psychological distress, social functioning impairment, and gastrointestinal/appetite-related symptoms. A low-burden group (80.2%) and a high-burden group (19.8%) were identified. The overall network's most central symptom was "reduced social interaction" (r = 2.52). Within groups, "worry" was most central in the low-burden group (r = 2.08), whereas "incision pain" was most central in the high-burden group (r=2.56). Computer simulations revealed that, within alleviating interventions, somatic symptoms accounted for the highest proportion across all symptom types (incision pain: -20.68% to -9.17%). Conversely, within aggravating interventions, psychosocial symptoms constituted the largest share in both the overall symptom network and the low-burden group (helplessness: 24.13% to 30.27%), whereas somatic symptoms dominated the high-burden group (incision pain: 9.17%).

A multimethod pipeline identified reduced social interaction, worry, and incision pain as profile-specific, structurally prominent symptoms within the postoperative day 3 symptom networks. These findings highlight distinct psychosocial and somatic features across symptom-burden profiles and provide hypotheses for subsequent longitudinal and interventional studies.