文献速递医学教育与科研方法流行病学与生物统计

一项系统综述描述临床试验设计阶段招募预测的模型

A systematic review describes models for recruitment prediction at the design stage of a clinical trial.
PubMed 文献 · J Clin Epidemiol · 2019-07-09

该研究系统检索2008至2016年文献及2008年前相关综述和灰色文献,纳入13篇文献,其中11篇聚焦随机方法、1篇聚焦确定性模型、1篇兼顾两者。模型在纳入因素和复杂性上差异较大,关键方面包括按时间条件化的能力、使用平均或中心特异性招募率以及中心启动率假设。部分模型缺乏灵活性限制了实施。

要点

  • 系统综述纳入13篇文献,11篇聚焦随机方法,1篇聚焦确定性模型,1篇兼顾两者。
  • 模型在纳入因素和复杂性上差异较大,关键方面包括时间条件化、招募率类型和中心启动率假设。
  • 确定性模型需指定少量参数但可能不现实,随机模型需更多参数且复杂性可能成为实施障碍。

核心结论

  • 确定性模型需要指定少量参数但可能不现实,尽管易于实施;随机模型日益需要更多参数指定,其复杂性可能成为实施障碍。

研究设计与样本

  • 系统综述,检索在线临床试验招募研究资源数据库(2008至2016年发表文章)、相关系统综述(2008年前发表文章)及谷歌搜索灰色文献,纳入13篇 eligible 文章。

主要发现

  • 13篇 eligible 文章中,11篇聚焦随机方法,1篇聚焦确定性模型,1篇兼顾随机和确定性方法。
  • 模型在纳入因素和复杂性上差异较大,关键方面包括按时间条件化的能力、使用平均或中心特异性招募率以及中心启动率假设。
  • 部分模型缺乏灵活性限制了实施。

局限性

  • 部分模型缺乏灵活性限制了实施。

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

OBJECTIVE: Patient recruitment in clinical trials is challenging with failure to recruit to time and target sample size common. This may be caused by unanticipated problems or by overestimation of the recruitment rate. This study is a systematic review of statistical models to predict recruitment at the design stage of clinical trials. STUDY DESIGN AND SETTING: The Online Resource for Recruitment research in Clinical triAls database was searched to identify articles published between 2008 and 2016. Articles published before 2008 were identified from a relevant systematic review. Google search was used to find potential methods in gray literature. RESULTS: Thirteen eligible articles were identified of which, 11 focused on stochastic approaches, one on deterministic models, and one included both stochastic and deterministic methods. Models varied considerably in the factors included and in their complexity. Key aspects included their ability to condition on time; whether they used average or center-specific recruitment rates; and assumptions around center initiation rates. Lack of flexibility of some models restricts their implementation. CONCLUSION: Deterministic models require specification of few parameters but are likely unrealistic although easy to implement. Increasingly, stochastic models require greater parameter specification, which, along with greater complexity may be a barrier to their implementation.

来源:J Clin Epidemiol,查看原文。版权归原作者所有。

引用

  1. Gkioni E, Rius R, Dodd S, et al. A systematic review describes models for recruitment prediction at the design stage of a clinical trial. J Clin Epidemiol. 2019. PMID: 31299358. doi:10.1016/j.jclinepi.2019.07.002

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。