Symptom networks of internalizing and externalizing problems in children with leukemia: a network analysis and...

Children with leukemia commonly experience internalizing and externalizing problems, yet their symptom-level interactions remain unclear. Network analysis provides a framework to examine associations among co-occurring symptoms and to explore potential intervention-relevant targets within a system-level context.

A total of 1,126 children with leukemia in the rehabilitation phase were included. Emotional and behavioral problems were assessed using the parent-reported Strengths and Difficulties Questionnaire (SDQ). Symptom endorsement rates were calculated, and an Ising network model based on binary data was estimated. Central and bridge symptoms were identified using expected influence metrics. Simulation analyses were conducted using the NodeIdentifyR algorithm (NIRA) to model symptom perturbations within the network. These simulation-based perturbations were applied in the overall network and internalizing-externalizing subnetworks to examine system-level responses.

"Motor fidgeting," "peer victimization," and "low mood" showed the highest centrality, with "motor fidgeting" and "peer victimization" also linking internalizing and externalizing problems. In simulation analyses, simulated alleviation of "impulsivity" was associated with the largest reduction in overall symptom burden (19.71%), whereas simulated aggravation of "motor fidgeting" showed the largest increase (22.76%). Externalizing-to-internalizing simulations showed patterns broadly consistent with the overall network structure. In internalizing-to-externalizing simulations, "low peer acceptance" showed the largest simulated reduction in externalizing symptoms (9.75%), whereas "low mood" showed the largest simulated increase (12.79%).

Internalizing and externalizing problems in children with leukemia form an interconnected symptom system. Simulation findings highlight symptom-specific patterns related to impulse control, peer relationships, and emotion regulation, which may inform early psychosocial monitoring and provide a basis for mechanism-informed future intervention strategies in pediatric oncology. These findings should be interpreted as hypothesis-generating evidence derived from cross-sectional data rather than as evidence of causal intervention effects.