Exploring the dynamics of perioperative symptom networks in colorectal cancer patients: a cross-lagged panel network analysis.

Background: Colorectal cancer incidence is on the rise, necessitating precise symptom management. However, causal relationships among symptoms have been challenging to establish due to reliance on cross-sectional data. Cross-lagged panel network (CLPN) analysis offers a solution, leveraging longitudinal data for insight.

Objective: We employed CLPN analysis to construct symptom networks in colorectal cancer patients at three perioperative time points, aiming to identify predictive relationships and intervention opportunities.

Methods: We evaluated the prevalence and severity of symptoms throughout the perioperative period, encompassing T1 the first day of admission, T2 2-3 days postoperatively, and T3 discharge, utilizing the M. D. Anderson Symptom Inventory Gastrointestinal Cancer Module (MDASI-GI). To identify crucial nodes in the network and explore predictive and interactive effects among symptoms, CLPNs were constructed from longitudinal data in R.

Results: The analysis revealed a stable network, with disturbed sleep exhibiting the highest out-EI (outgoing expected influence) during T1. Distress had a sustained impact throughout the perioperative. Disturbed sleep at T1 predicted T2 bloating, fatigue, distress, and pain. T1 distress predicted T2 sadness severity. T2 distress primarily predicted T3 fatigue, disturbed sleep, changes in taste, and bloating. T2 shortness of breath predicted T3 changes in taste and loss of appetite. Furthermore, biochemical markers like RBC and ALB had notable influence on symptom clusters during T1→T2 and T2→T3, respectively.

Conclusion: Prioritizing disturbed sleep during T1 and addressing distress throughout the perioperative phase is recommended. Effective symptom management not only breaks the chain of symptom progression, enhancing healthcare impact, but also eases patient symptom burdens.