腺性膀胱炎复发风险预测模型的建立和验证

目的 探讨并分析腺性膀胱炎（CG）初治后复发的主要临床危险因素，并构建定量预测模型指导临床早期识别高危患者。方法 回顾性总结2015年8月至2020年7月在本院经病理确诊的138例CG患者的临床资料，采用随机数字表法分为模型组（92例）和验证组（46例），以病理和临床症状联合诊断为复发共62例，其中模型组43例（46.7%）和验证组19例（41.3%）。首先采用单因素分析比较模型组中复发与未复发患者的临床资料，然后采用多因素Cox比例风险模型筛选主要危险因素，根据权重赋值建立定量预测模型，采用受试者工作特征（ROC）曲线分析预测模型在模型组和验证组中的诊断效能，最后根据预测模型将模型组和验证组进行风险分层（低、中和高），采用Kaplan-Meier曲线比较复发风险。结果 模型组的单因素分析结果显示，与未复发患者比较，复发患者的年龄更大，留置尿管时间延长，尿路感染、尿路结石、鳞状上皮化生和非典型增生的发生率升高（均P<0.05）。Cox分析结果显示，年龄、留置尿管时间、尿路感染、尿路结石、鳞状上皮化生和非典型增生是CG复发的主要危险因素（均P<0.05）。ROC曲线分析结果显示，预测模型在模型组和验证组中诊断复发的准确性分别为0.876和0.845（P<0.001）。Kaplan-Meier曲线分析结果显示，模型组和验证组的高风险患者的复发风险高于中风险患者，中风险患者的复发风险高于低风险患者（均P<0.05）。结论 CG有较高的病理和临床症状复发率，年龄、留置尿管时间、尿路感染、尿路结石、鳞状上皮化生和非典型增生可能是CG复发的主要危险因素，通过构建定量预测模型对诊断CG复发和风险分层有较高的准确性和应用价值。

Establishment and validation of a recurrence risk prediction model for cystitis glandularis

Objective  To investigate and analyze the main clinical risk factors of cystitis glandularis (CG) recurrence after initial treatment, and construct a quantitative prediction model to guide clinical early identification of high risk patients.Methods  Clinical date of a total of 138 patients with CG confirmed by pathology in our hospital from August 2015 to July 2020 were retrospectively summarized. They were randomly divided into model group (92 cases) and verification group (46 cases ) . Sixty-two patients of recurrence were diagnosed by pathological and clinical symptoms, including 43 patients in model group (46.7%) and 19 patients (41.3%) in validation group. Firstly, the clinical datas between recurrence and non-recurrence patients in model group were compared by single factor, then the main risk factors were screened by multivariate Cox proportional hazard model, and the quantitative prediction model was established according to the weight assignment. The receiver operating characteristic (ROC)curve was used to analyze the diagnostic efficacy of the prediction model in model group and validation group. Finally, the risk of model group and validation group was evaluated according to the prediction model,and Kaplan Meier curve was used to compare the recurrence risk during the low,median and high risk patients.Results  Univariate analysis of the model group showed that compared with the patients without recurrence, the patients with recurrence had higher age, longer indwelling catheter time, and higher incidence of urinary tract infection, urinary calculi, squamous metaplasia and atypical hyperplasia (all P<0.05). Cox analysis showed that age, indwelling catheter time, urinary tract infection, urinary calculi, squamous metaplasia and atypical hyperplasia were the main risk factors for CG recurrence (all P<0.05). ROC curve analysis showed that the accuracy of the prediction model in the diagnosis of recurrence in the model group and the validation group was 0.876 and 0.845, respectively (P<0.001). Kaplan-Meier curve analysis showed that the recurrence risk of medium-high risk patients was higher than that of medium-risk patients in the model group and validation group, and the recurrence risk of medium risk patients was higher than that of low-risk patients (all P<0.05).Conclusions  CG has a high recurrence rate of pathological and clinical symptoms. Age, indwelling catheter time, urinary tract infection, urinary calculi, squamous metaplasia and atypical hyperplasia may be the main risk factors to CG recurrence. The establishment of quantitative prediction model has high accuracy and application value for the diagnosis and risk stratification of recurrence.