可改变的方法学与报告实践与健康科学研究可重复性相关:一项系统综述与证据缺口图
该研究通过系统检索五个文献数据库(2000年1月至2023年5月),纳入148项报告可重复性相关因素的健康科学研究,采用证据缺口图汇总分析。结果显示,样本量与检验效能、分析方法和报告透明度等可改变因素常与更好的可重复性相关;作者建议优先采用更严格的统计检验程序与阈值、进行适当的样本量与检验效能计算,并提高报告的透明度与完整性。
要点
- 样本量与检验效能、分析方法和报告透明度等可改变因素与更好的可重复性相关。
- 在234项因素评估中,155项(66%)进行了与可重复性结局的统计学关联检验。
- 作者建议优先采用更严格的统计检验程序与阈值,并提高报告透明度与完整性。
核心结论
- 若干可改变因素与健康科学研究可重复性相关,为干预提供了机会。
- 应用更严格的统计检验程序与阈值、进行适当的样本量与检验效能计算、提高报告的透明度与完整性应是改善可重复性的优先事项。
- 需要开展实验性研究来检验改善可重复性的干预措施。
研究设计与样本
- 系统综述与证据缺口图,检索五个文献数据库2000年1月至2023年5月的记录,并补充检索高影响力期刊与相关记录。
- 纳入报告可重复性相关因素数据的观察性、干预性或知识合成研究的健康科学记录,共纳入148项研究。
- 纳入研究主要来自生物医学/临床前领域(n=62)和临床领域(n=71)。
- 采用归纳定性内容分析编码数据,并以证据缺口图合成实证证据。
- 因素分为可改变与不可改变两类,可重复性结局分为方法与结果两类。
主要发现
- 因素被归类为12类可改变因素(如样本量与检验效能)和3类不可改变因素(如发表年份)。
- 在234项报告的因素评估中,76项(32%)评估方法学可重复性,158项(68%)评估结果可重复性。
- 报告最多的因素为透明度与报告(234项评估中38项)。
- 共155项因素(66%)评估了与可重复性结局的统计学关联。
- 统计学关联检验最常针对分析方法(26项报告显著性中24项)、样本量与检验效能(23项报告显著性中21项)、参与者特征与研究材料(12项报告显著性中10项)。
临床怎么用
- 作者建议:应用更严格的统计检验程序与阈值、进行适当的样本量与检验效能计算、提高报告的透明度与完整性应是改善可重复性的优先事项。
- 作者认为需要开展实验性研究来检验改善可重复性的干预措施。
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: Map the evidence on factors (eg, research practices) associated with reproducibility of methods and results reported in health sciences research. STUDY DESIGN AND SETTING: Five bibliographic databases were searched from January 2000 to May 2023 with supplemental searches of high-impact journals and relevant records. We included health science records of observational, interventional, or knowledge synthesis studies reporting data on factors related to research reproducibility. Data were coded using inductive qualitative content analysis, and empirical evidence was synthesized with evidence and gap maps. Factors were categorized as modifiable or nonmodifiable; reproducibility outcomes were categorized as related to methods or results. Statistical tests of association between factors and reproducibility outcomes were summarized. RESULTS: Our review included 148 studies, primarily from biomedical/preclinical (n = 62) and clinical (n = 71) domains. Factors were classified into 12 modifiable (eg, sample size and power) and three nonmodifiable (eg, publication year) categories. Of 234 reported evaluations of factors, 76 (32%) assessed methodological reproducibility and 158 (68%) assessed results reproducibility. The most frequently reported factor was transparency and reporting (38 of 234 assessments). A total of 155 factors (66%) were evaluated for statistical associations with reproducibility outcomes. Statistical associations were most frequently conducted for analytical methods (24 of 26 reporting significance), sample size and power (21 of 23 reporting significance), and participants characteristics and study materials (10 of 12 reporting significance). CONCLUSION: Several modifiable factors were associated with reproducibility of health sciences research and represent opportunities for intervention. Applying more stringent statistical testing procedures and thresholds, conducting appropriate sample size and power calculations, and improving transparency and completeness of reporting should be top priorities for improving reproducibility. Experimental studies to test interventions to improve reproducibility are needed. PLAIN LANGUAGE SUMMARY: Many research findings in medicine and health cannot be reproduced by other researchers. This makes it harder to know what evidence to trust when making patient care and health policy decisions. We systematically reviewed 148 studies that examined how specific research practices are linked to whether research findings can be reproduced. We found that several modifiable features of study design, analysis, and reporting were often associated with better reproducibility. Using larger sample sizes, applying more stringent statistical methods, and providing transparent, complete descriptions of methods and results were frequently linked to more reproducible findings. These results suggest that improving how studies are planned, analyzed, and reported may increase the trustworthiness and usefulness of health research.
引用
- Kennett J, Moss SJ, Parsons Leigh J, et al. Modifiable methodological and reporting practices are associated with reproducibility of health sciences research: a systematic review and evidence and gap map. J Clin Epidemiol. 2026. PMID: 41519488. doi:10.1016/j.jclinepi.2026.112135
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。