预测降尿酸治疗期间痛风频繁发作的临床模型建立与评价

Establishment and Evaluation of a Clinical Model to Predict Frequent Gout Attacks during Urate-Lowering Therapy

摘要: 目的：确定降尿酸治疗过程中影响痛风患者痛风频繁发作风险的因素。方法：本研究为单中心、观察性研究，纳入2022年8月至2023年8月在青岛大学附属医院痛风门诊初诊且确诊为痛风的患者165名。入组后，所有患者开始降尿酸治疗(ULT)，并在3个月、6个月、12个月时接受随访。在12个月内出现至少2次急性发作的患者(频繁痛风发作患者)与小于两次急性发作的患者(非频繁痛风发作患者)进行比较，计算痛风发作风险的比值比(ORs)。利用多因素logistic回归建立临床预测模型，并绘制受试者工作曲线(ROC)评价模型预测性能。结果：111名患者被归类为频繁痛风发作患者，54名患者被归类为非频繁痛风发作患者。痛风频繁发作患者的血沉、肉眼痛风石、痛风家族史较非频繁发作患者显著增高(P < 0.05)。两组患者在是否患有高血压、是否患有脂肪肝、年龄、尿酸和饮酒史方面差异无统计学意义(P > 0.05)。结论：本研究建立的预测痛风频繁发作预测模型的预测效能较好，为临床及时识别频繁痛风发作患者提供了一个参考工具。

Abstract: Objective: To determine the factors that affect the risk of frequent gout attacks in gout patients during uric acid-lowering treatment. Methods: This study is a single-center, observational study, including 165 patients who were first diagnosed with gout at the Gout Clinic of the Affiliated Hospital of Qingdao University from August 2022 to August 2023. After enrollment, all patients started urate-lowering therapy (ULT) and were followed up at 3 months, 6 months, and 12 months. Odds ratios (ORs) were calculated for the risk of gout attacks in patients who had at least two exacerbations within 12 months (frequent gout attack patients) compared with those who had less than two exacerbations (infrequent gout attack patients). Multi-factor logistic regression was used to establish a clinical prediction model, and a receiver operating curve (ROC) was drawn to evaluate the prediction performance of the model. Results: 111 patients were classified as having frequent gout attacks, and 54 patients were classified as having infrequent gout attacks in remission. The erythrocyte sedimentation rate, gross tophi, and family history of gout in patients with frequent gout attacks were significantly higher than those in the non-frequent attack group (P < 0.05). There was no statistically significant difference between the two groups in terms of whether they had hypertension, fatty liver, age, uric acid and drinking history (P > 0.05). Conclusion: The prediction model established in this study for predicting frequent gout attacks has good prediction performance and provides a reference tool for clinical timely identification of patients with frequent gout attacks.