目的 探讨癌症患者应用羟考酮镇痛的效果及其影响因素，构建镇痛效果预测模型，为临床个体化镇痛方案的制定提供参考。方法 回顾性收集使用羟考酮患者的临床资料，包括性别、年龄、住院天数、羟考酮剂量、癌种、是否转移、转移部位、病理分级、TNM分期、合并疾病、体表面积（BSA）、疼痛部位及疼痛类别等指标，以镇痛缓解率为效应指标。通过单因素分析和二元Logistic回归模型确定影响因素，并构建列线图预测模型，用受试者工作特征（ROC）曲线和曲线下面积（AUC）评估模型性能。采用决策曲线分析（DCA）量化模型在不同阈值概率下的净获益。结果 共纳入284例癌症患者，羟考酮镇痛缓解率为74.3%。单因素分析显示，年龄、BSA、癌种、TNM分期、疼痛类别、疼痛部位及羟考酮剂量与镇痛效果相关（P＜0.05）；二元Logistic回归分析表明，年龄（P＜0.001）、BSA（P＝0.024）、上腹部疼痛（P＝0.001）、胀痛（P＝0.016）、TNM分期（I/II期）（P＜0.001）及羟考酮剂量（P＝0.011）是羟考酮镇痛效果的影响因素。基于上述因素构建的列线图模型AUC为0.922[95%置信区间（CI）： 0.892～0.955]，校准曲线显示预测值与实际值一致性良好，表明该模型具有较好的预测价值。DCA表示，在临床常用的阈值概率区间（0.2～0.7）内，应用该预测模型指导羟考酮镇痛疗效的个体化预测，可获得额外的临床净获益，具有良好的临床实用价值。结论 研究通过统计分析，确定了年龄、BSA、肿瘤类型、TNM分期、疼痛类型、疼痛部位以及羟考酮剂量作为影响羟考酮镇痛效果的关键变量。基于这些变量构建的列线图预测模型展现了优异的预测效能，为临床医师制定个性化镇痛治疗方案提供了有力的辅助工具。

Objective To explore the effect and influencing factors of oxycodone analgesia in cancer patients, construct the prediction model of analgesic effect, and provide reference for the formulation of individualized analgesic scheme in clinical practice. Methods The clinical data of patients using oxycodone were retrospectively collected, including gender, age, length of hospital stay, oxycodone dosage, type of cancer, whether metastasis occurred, metastatic sites, pathological grade, TNM stage, comorbidities, body surface area (BSA), pain location and pain category, etc. The analgesic relief rate was used as the effect indicator. Univariate analysis and binary Logistic regression model were used to determine the influencing factors, and a nomogram prediction model was constructed. The performance of the model was evaluated by ROC curve and AUC. The net benefit was quantified by decision curve analysis (DCA) model under different threshold probabilities. Results A total of 284 cancer patients were included in the study, with a hydrocodone analgesic response rate of 74.3%. Univariate analysis revealed that age, BSA, cancer type, TNM stage, pain category, pain location, and hydrocodone dosage were significantly associated with analgesic efficacy (P < 0.05). Binary logistic regression analysis identified the following factors as predictors: age (P < 0.001), BSA (P = 0.024), upper abdominal pain (P = 0.001), distending pain (P = 0.016), TNM stage (I/II) (P < 0.001), and hydrocodone dosage (P = 0.011). The constructed logistic regression model demonstrated an AUC of 0.922 (95% CI: 0.892—0.955), with calibration curves showing good agreement between predicted and actual values, indicating strong predictive value. DCA indicates that within the clinically commonly used threshold probability interval (0.2—0.7), applying this prediction model to guide individualized prediction of oxycodone analgesic efficacy can yield additional clinical net benefits, demonstrating significant clinical utility. Conclusion This study identified age, BSA, tumor type, TNM staging, pain type, pain location, and oxycodone dose as key variables affecting oxycodone's analgesic efficacy through statistical analysis. The nomogram model developed based on these variables demonstrated excellent predictive performance, providing clinicians with a powerful tool for developing personalized pain management plans.

R971

国家自然科学基金资助项目（81872236）；天津市医学重点学科建设项目（TJYXZDXK-3-007B）