| 摘要: |
| 目的 系统性评价前列腺癌根治术后尿失禁风险预测模型的性能和方法学质量，为临床医护人员选择合适的风险评估工具提供参考。方法 利用PubMed、Web of Science、Cochrane Library、CINAHL、EMBASE、中国知网、万方、维普、中国生物医学文献数据库，检索建库至2024年1月23日发表的相关文献。由2名研究者独立进行文献筛选、资料提取，并应用预测模型偏倚风险工具评估所纳入研究的偏倚风险和适用性。采用MedCalc软件，使用随机效应模型对验证组的AUC进行meta分析，并进行发表偏倚评估和敏感性分析。结果 共纳入8项研究，样本量共7 216例，6个模型报告了AUC值，7个模型报告了校准度。2项研究的适用性较好，6项研究的适用性较差。最常用的预测模型类型为logistic回归模型，剔除极端AUC值的模型后，随机效应meta分析结果为0.840（95% CI 0.786～0.895），异质性检验I2＝0%（P＝0.737）。8项研究的偏倚风险均较高，偏倚来源主要为数据来自回顾性队列研究、部分连续性变量转化为二分类变量、缺失数据未处理、基于单因素分析筛选预测因子、未完整报告模型的区分度和校准度及缺乏模型的外部验证等。Egger检验结果提示研究无显著发表偏倚。结论 现有前列腺癌根治术后尿失禁风险预测模型的开发和验证过程尚不完善，未来研究应构建基于多中心、大样本数据的风险预测模型，加强对模型的临床适用性评估，并严格遵循预测模型报告规范与流程，从而建立可用于临床实践的高质量风险预测模型。 |
| 关键词: 风险预测模型 根治性前列腺切除术 尿失禁 系统评价 meta分析 |
| DOI：10.16781/j.CN31-2187/R.20240346 |
| 投稿时间：2024-05-22修订日期：2024-06-25 |
| 基金项目:国家自然科学基金（81903182），海军军医大学深蓝护理科研项目（2022KYD09）． |
|
| Risk prediction models for urinary incontinence after radical prostatectomy: a systematic review and meta-analysis |
| GAO Chengfei,GU Jie,ZHANG Wenhui,GAO Pei,GAO Xu*,CAO Jie* |
(Department of Urology, The First Affiliated Hospital of Naval Medical University (Second Military Medical University), Shanghai 200433, China
*Corresponding authors) |
| Abstract: |
| Objective To systematically evaluate the performance and methodological quality of the risk prediction models for urinary incontinence after radical prostatectomy, so as to provide a reference for selecting the appropriate risk prediction tool. Methods A systematic search was conducted in PubMed, Web of Science, Cochrane Library, CINAHL, EMBASE, CNKI, Wanfang, VIP, and Chinese biomedical literature database from inception to Jan. 23, 2024. Two researchers independently conducted literature screening and data extraction, and the prediction model risk of bias assessment tool (PROBAST) was applied to assess the risk of bias and applicability of the included studies. MedCalc software was used to perform a meta-analysis of the area under curve (AUC) of the validation groups using the random effect model, and the publication bias and sensitivity analysis were also performed. Results A total of 8 studies were included, with a combined sample size of 7 216 cases. Six models reported the AUC values, and 7 models reported calibration. The applicability of 2 studies was acceptable, while 6 were poor. The most commonly used type of prediction model was logistic regression. After excluding models with extreme AUC values, the random-effects meta-analysis result was 0.840 (95% confidence interval 0.786 to 0.895), with no heterogeneity (I2＝0%, P＝0.737). The bias risk was high in all 8 studies, mainly due to retrospective cohort data, transformation of continuous variables into binary variables, unaddressed missing data, selection of predictors based on univariate analysis, incomplete report of the model discrimination and calibration, and lack of external validation. Egger test result indicated no significant publication bias. Conclusion The development and validation process of the existing risk prediction models for urinary incontinence after radical prostatectomy is still imperfect. Future research should construct prediction models based on multicenter and large-sample data, strengthen the clinical applicability assessment of the models, and strictly follow the reporting standards and procedures, so as to establish high-quality risk prediction models for clinical practice. |
| Key words: risk prediction models radical prostatectomy urinary incontinence systematic review meta-analysis |
