带状疱疹后神经痛共患抑郁障碍的影响因素分析及预测模型构建

Analysis of risk factors and development of a prediction model for depressive disorder following postherpetic neuralgia

摘要:

目的：探讨带状疱疹后神经痛（Post-herpetic Neuralgia, PHN）患者共患抑郁障碍的影响因素，并构建预测模型。方法：回顾性分析2021年03月至2024年12月期间福建医科大学附属第一医院疼痛科收治的282例PHN患者的临床资料，根据患者是否共患抑郁障碍分为PHN共患抑郁障碍组（n = 94）和单纯PHN组（n = 188），通过Logistic回归筛选其影响因素，构建列线图预测模型并综合评价模型性能。结果：多因素Logistic回归分析显示PHN病程、恶性肿瘤史、NRS评分、受教育程度、SSRS评分是PHN共患抑郁障碍的影响因素。通过Bootstrap验证法进行内部验证后，列线图预测PHN共患抑郁障碍的AUC值为0.970（95% CI：0.941 - 0.987），灵敏度0.890，特异度0.941，显示模型区分能力可靠；Hosmer-Lemeshow检验结果显示X2 = 5.293，P = 0.726，Brier评分为0.049，校准曲线显示模型具有较好的校准度及预测一致性；临床决策曲线分析（Decision Curve Analysis, DCA）显示模型具有良好的临床适用性。结论：PHN病程、恶性肿瘤史、NRS评分是PHN共患抑郁障碍的独立危险因素，受教育程度、SSRS评分是其独立保护因素，据此构建的列线图模型具有可靠的临床预测价值。

Abstract:

Objective: Investigation of influencing factors and development of a prediction model for comorbid depressive disorder in patients with postherpetic neuralgia. Methods: A retrospective analysis was conducted on clinical data from 282 patients with postherpetic neuralgia admitted to the Department of Pain Medicine at the First Affiliated Hospital of Fujian Medical University between March 2021 and December 2024. Based on the presence of comorbid depressive disorder, patients were divided into two groups: PHN with comorbid depression (n = 94) and PHN alone (n = 188). Logistic regression analysis was employed to identify influencing factors, followed by the construction of a nomogram prediction model. The performance of the model was comprehensively evaluated. Results: Multivariate logistic regression analysis identified duration of PHN, history of malignant tumor, NRS score, educational attainment, and SSRS score as significant influencing factors for comorbid depressive disorder in PHN. Following internal validation via the bootstrap method, the nomogram achieved an AUC of 0.970 (95% CI: 0.941 - 0.987) for predicting comorbid depression, with a sensitivity of 0.890 and specificity of 0.941, indicating excellent discriminative ability. The Hosmer-Lemeshow test result shows that X2 = 5.293, P = 0.726, and the Brier score is 0.049. The calibration curve indicates that the model has good calibration and prediction consistency. The clinical decision curve analysis demonstrated that the model has good clinical applicability. Conclusion: Duration of PHN, history of malignant tumor, and NRS score were identified as independent risk factors for comorbid depressive disorder in PHN, while educational attainment and SSRS score served as independent protective factors. The nomogram prediction model constructed based on these factors demonstrates robust clinical predictive value.