Symptom network in patients with colorectal cancer during the perioperative period.

[PURPOSE] This study employed cross-lagged panel network (CLPN) analysis to construct a symptom network for colorectal cancer patients at four time points during the perioperative period, aiming to identify predictive relationships and intervention opportunities.

[METHODS] A total of 315 patients who underwent colorectal cancer surgery were enrolled in this study. They were assessed at four time points: preoperatively at admission (T0), and postoperatively on day 1 (T1), day 3 (T2), and on the day of discharge (T3). Symptom severity was evaluated using the MD Anderson Symptom Scale and the colorectal cancer module. Cross-sectional networks were analyzed using Gaussian graphical models (GGM), while CLPN models with LASSO regularization examined directed temporal relationships between symptoms, adjusting for covariates.

[RESULTS] The analysis revealed dynamic changes in core symptoms: vomiting (T0-EI = 1.090), distress (T1/T2-EI > 1.021), and altered bowel habits (T3-EI = 0.461). Overall network strength significantly decreased from T2 to T3 (p < 0.001). In the T0-T1 band, sadness at admission (OEI = 0.684) predicted vomiting (edge weight = 0.297), low mood (0.183), and dry mouth (0.088) on postoperative day 1; in the T1-T2 band, low mood at T1 (OEI = 0.651) predicted altered stool characteristics (0.502) and constipation (0.370) at T2. Vomiting and altered stool characteristics showed incoming predictability of 0.41 and 0.45, respectively.

[CONCLUSION] Preoperative emotional symptoms play a key predictive role in subsequent gastrointestinal symptoms. By intervening in psychological stress and intestinal dysfunction before and immediately after surgery, it may be possible to interrupt the symptom transmission pathway. Symptom network analysis provides actionable targets for precision intervention.