Background: Colorectal cancer (CRC) is common in China, and many postoperative patients experience fear of cancer recurrence (FCR), which negatively affects mental health and quality of life. Predictors of FCR remain underexplored, and traditional regression may overlook complex interactions, whereas random forest (RF) modeling allows robust variable selection.
Purpose: To assess FCR prevalence and identify associated factors using a RF model.
Methods: Between November 2023 and May 2024, 314 postoperative colorectal cancer patients were enrolled at Fudan University Shanghai Cancer Center. Data were collected using the Brief Illness Perception Questionnaire (BIPQ), Cognitive Emotion Regulation Questionnaire (CERQ), Social Support Rating Scale (SSRS), and Social Constraints Scale-15 (SCS-15). Predictors were ranked using a random forest model and confirmed via binary logistic regression.
Results: High FCR was reported by 58.9% of patients. Key predictors (> 5%) included illness perception, negative and positive emotion regulation, social support, social constraints, age, and income. Univariate analyses showed strong associations for BIPQ, CERQ-maladaptive (CERQ-M), SCS-15 (P < 0.001), with income, SSRS, and CERQ-adaptive (CERQ-A) also significant (P < 0.05). Logistic regression confirmed BIPQ, SCS-15, and CERQ-M predicted FCR. Age, despite > 5%, was not independently significant, likely due to adjustment for other psychosocial factors and complex interactions.
Conclusion: A substantial proportion of CRC survivors experience high FCR. Illness perceptions, social constraints, and maladaptive coping are key determinants. Integrating psychosocial screening and targeted interventions into postoperative care may reduce FCR and improve quality of life.
Keywords: CRC; Determinants; FCR; RF algorithm.
© 2025. The Author(s).