支持乳腺癌决策的922个预测模型多数存在高偏倚风险
该系统综述检索2010年1月至2020年12月文献,筛选20460项研究后纳入534项研究、922个模型,用PROBAST评估偏倚风险。模型预测死亡(417个,45%)、复发(217个,24%)等结局,285个模型(31%)缺乏完整描述无法应用,878个模型(95%)存在高偏倚风险。作者认为需进一步开发经充分质量与效度评估的预测模型。
要点
- 922个模型中878个(95%)经PROBAST评估存在高偏倚风险。
- 285个模型(31%)缺乏完整最终模型描述,无法用于新患者。
- 模型间预测变量与结局存在大量重叠,多数未按既定报告指南报告。
核心结论
- 多数模型未按既定报告指南报告,或在开发与/或验证中存在方法学缺陷。
- 进一步开发经充分质量与效度评估的预测模型是未来临床应用必要的第一步。
研究设计与样本
- 系统综述,系统检索2010年1月至2020年12月发表的、报告旨在支持乳腺癌治疗决策的预测模型开发的研究。
- 筛选20460项研究后纳入534项研究,报告922个模型,使用PROBAST评估质量与偏倚风险。
主要发现
- 922个模型预测:死亡417个(45%)、复发217个(24%)、淋巴结受累141个(15%)、不良事件58个(6%)、治疗反应56个(6%)、其他结局33个(4%)。
- 285个模型(31%)缺乏完整最终模型描述,无法应用于新患者。
- 878个模型(95%)被认为存在高偏倚风险。
- 观察到模型间预测变量与结局存在大量重叠。
临床怎么用
- 原文未据此给出推荐。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVES: To systematically review the currently available prediction models that may support treatment decision-making in breast cancer. STUDY DESIGN AND SETTING: Literature was systematically searched to identify studies reporting on development of prediction models aiming to support breast cancer treatment decision-making, published between January 2010 and December 2020. Quality and risk of bias were assessed using the Prediction model Risk Of Bias (ROB) Assessment Tool (PROBAST). RESULTS: After screening 20,460 studies, 534 studies were included, reporting on 922 models. The 922 models predicted: mortality (n = 417 45%), recurrence (n = 217, 24%), lymph node involvement (n = 141, 15%), adverse events (n = 58, 6%), treatment response (n = 56, 6%), or other outcomes (n = 33, 4%). In total, 285 models (31%) lacked a complete description of the final model and could not be applied to new patients. Most models (n = 878, 95%) were considered to contain high ROB. CONCLUSION: A substantial overlap in predictor variables and outcomes between the models was observed. Most models were not reported according to established reporting guidelines or showed methodological flaws during the development and/or validation of the model. Further development of prediction models with thorough quality and validity assessment is an essential first step for future clinical application.
引用
- Hueting TA, van Maaren MC, Hendriks MP, et al. The majority of 922 prediction models supporting breast cancer decision-making are at high risk of bias. J Clin Epidemiol. 2022. PMID: 36633901. doi:10.1016/j.jclinepi.2022.10.016
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。