P 值和贝叶斯分析在过去 10 年（2007 年至 2017 年）发表的儿童健康研究随机对照试验中的应用：方法学综述方案。

P value and Bayesian analysis in randomized-controlled trials in child health research published over 10 years, 2007 to 2017: a methodological review protocol.

机构信息

The Children's Hospital Research Institute of Manitoba, 513-715 McDermot Avenue, John Buhler Research Centre, Winnipeg, Manitoba, R3E 3P4, Canada.

Department of Pediatrics and Child Health, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, MB, Canada.

出版信息

Syst Rev. 2021 Mar 10;10(1):71. doi: 10.1186/s13643-021-01622-8.

DOI:10.1186/s13643-021-01622-8

PMID:33691775

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC7948362/

Abstract

BACKGROUND

There is an unresolved debate about the reliability of the interpretation of P value. Some investigators have suggested that an alternative Bayesian method is preferred in conducting health research. As randomized-controlled trials (RCTs) are important in generating research evidence, we decided to investigate the extent, if any, the inferential statistical framework in published RCTs in child health research have changed over 10 years. We aim to examine the change in P value and Bayesian analysis in RCTs in child health research papers published from 2007 to 2017.

METHODS

We will search the Cochrane Central Register of Controlled Trials (Wiley) to identify relevant citations. We will leverage a pre-existing sample of child health RCTs published in 2007 (n=300) used in our previous study of reporting quality of pediatric RCTs. Using the same strategy and study selection methods, we will identify a comparable random sample of child health RCTs published in 2017 (n=300). Eligible studies will include RCTs in health research among individuals aged 21 years and below. One reviewer will select studies for inclusion and extract the data and another reviewer will verify these. Disagreements will be resolved by a discussion between reviewers or by involving another reviewer. We will perform a descriptive analysis of 2007 and 2017 samples and analyze the results using both the frequentist and Bayesian methods. We will present specific characteristics of the clinical trials from 2007 and 2017 in tabular and graphical forms. We will report the difference in the proportion of P value and Bayesian analysis between 2007 and 2017 to assess the 10-year change. Clustering around P values of significance, if observed, will be reported.

DISCUSSION

This review will present the difference in the proportion of trials that reported on P value and Bayesian analysis between 2007 and 2017 to assess the 10-year change. The implications for future clinical research will be discussed and this research work will be published in a peer-reviewed journal. This review has the potential to help inform the need for a change in the methodological approach from the null hypothesis significance test to Bayesian methods.

SYSTEMATIC REVIEW REGISTRATION

Open Science Framework https://osf.io/aj2df.

摘要

背景

关于 P 值解释的可靠性存在尚未解决的争议。一些研究人员建议，在进行健康研究时，首选替代的贝叶斯方法。由于随机对照试验（RCT）在产生研究证据方面很重要，我们决定调查过去 10 年来，儿童健康研究中已发表的 RCT 中推论统计框架是否有任何变化。我们旨在研究 2007 年至 2017 年发表的儿童健康研究论文中 P 值和贝叶斯分析的变化。

方法

我们将在 Cochrane 对照试验中央注册库（威利）中搜索相关引文。我们将利用我们之前关于儿科 RCT 报告质量的研究中使用的 2007 年发表的儿童健康 RCT 的现有样本（n=300）。使用相同的策略和研究选择方法，我们将确定 2017 年发表的儿童健康 RCT 的可比随机样本（n=300）。合格的研究将包括年龄在 21 岁及以下的个体的健康研究中的 RCT。一名评审员将选择纳入的研究并提取数据，另一名评审员将对这些数据进行验证。如果存在分歧，将通过评审员之间的讨论或邀请另一名评审员来解决。我们将对 2007 年和 2017 年样本进行描述性分析，并使用频率论和贝叶斯方法分析结果。我们将以表格和图形形式呈现 2007 年和 2017 年临床试验的具体特征。我们将报告 2007 年和 2017 年之间 P 值和贝叶斯分析报告比例的差异，以评估 10 年的变化。如果观察到围绕显著性 P 值的聚类，将报告聚类。