Dynamic symptom-burden trajectories and cross-lagged symptom networks after oesophagectomy: A multicentre longitudinal study

Purpose To clarify the recovery process among patients with oesophageal cancer and the patterns of association between symptoms. Methods A total of 887 patients undergoing oesophagectomy were recruited from four tertiary hospitals. Symptom burden was assessed on postoperative day 3, day 7, day 14, and at one month. Latent class growth modelling identified symptom-burden trajectories. Regularized cross-lagged panel networks examined lagged symptom-to-symptom predictive associations across adjacent recovery stages. Results Two trajectories were identified: a high-stable symptom burden trajectory (29.5%) and a low-decreasing trajectory (70.5%). Reduced social interaction showed the highest cross-lagged out-strength in the mean network (0.840) and the highest out-expected influence across all stages (0.503, 0.529, and 1.490). In the high-stable group, cross-lagged edges increased from 19 to 28 and 36, with reduced social interaction increasingly connected to subsequent physical symptoms in the later stage. In the low-decreasing group, cross-lagged edges were 24, 19, and 24, with early and middle associations mainly among psychological symptoms. Conclusion Postoperative symptoms following oesophagectomy showed distinct trajectories, and stage-specific predictive patterns. These findings support trajectory-informed risk stratification and timing-specific symptom management after oesophagectomy.