The construction and validation of a clinical predictive model for somatic symptom disorders in epilepsy patients

Objective To investigate factors influencing somatic symptom disorder (SSD) in epilepsy patients and construct a cut-off point prediction model. Methods Using structured interviews and based on DSM-5 diagnostic criteria, the 206 epilepsy patients included in this study were categorized into SSD and non-SSD (n-SSD) groups. Demographic and clinical data were collected, and assessments were conducted using the Quality of Life in Epilepsy (QOLIE-31), Generalized Anxiety Disorder-7 (GAD-7), Neuropsychiatric Disease and Disability Inventory-Extended (NDDI-E), and Pittsburgh Sleep Quality Index (PSQI). Age, negative life events, seizure anxiety, energy/fatigue, GAD-7, NDDI-E, and PSQI scores were identified as independent risk factors for SSD comorbidity in epilepsy. The constructed cut-off model demonstrated good predictive performance. External validation in an independent multicenter cohort is required prior to clinical implementation. Results Compared with the n-SSD group, the SSD group exhibited statistically significant differences in age, age at onset, years of education, place of residence, number of comorbid physical illnesses, and adverse life events (all p < 0.05). The SSD group also showed significantly higher scores on GAD-7, NDDI-E, and PSQI, but lower total QOLIE-31 score and lower subscale scores for seizure worry, medication effects, energy/fatigue, life satisfaction, social functioning, and emotional well-being (all p < 0.05). Multivariate logistic regression analysis revealed that age (OR = 1.076, 95% CI: 1.015–1.141), negative life events(OR = 6.624, 95% CI: 2.130–20.606), seizure anxiety (OR = 0.945, 95% CI: 0.895–0.999), energy/fatigue (OR = 0.923, 95% CI: 0.872–0.977), and GAD-7 (OR = 1.274, 95% CI: 1.015–1.274) were independently associated with higher QOLIE-31 total scores. Fatigue (OR = 0.923, 95% CI: 0.872–0.977), GAD-7 (OR = 1.274, 95% CI: 1.037–1.565), NDDI-E (OR = 1.233, 95% CI: 1.038–1.442), and PSQI (OR = 1.375, 95% CI: 1.097–1.723) were independent predictors of SSD. The AUC of the nomogram model constructed based on the aforementioned factors was 0.939 (95% CI: 0.904–0.975), with an AUC of 0.907 following internal validation. The optimal risk probability cutoff value was 0.200 (based on the Yorden index), yielding a sensitivity of 84.7% and specificity of 95.3%. Calibration curve and decision curve analyses demonstrated good model calibration and clinical net benefit. Conclusion Older age, exposure to negative life events, higher GAD-7, NDDI-E, and PSQI scores, and lower scores on the seizure worry and energy/fatigue dimensions of QOLIE-31 are independent risk factors for SSD in epilepsy patients. The constructed nomogram model demonstrates favorable predictive performance. External validation within an independent multicenter cohort is required prior to clinical implementation.