SYRCLE偏倚风险评估工具发布是否提高动物实验方法质量？

Does the publication of SYRCLE’s risk of bias tool improve the methodological quality of animal experiments?

目的基于SYRCLE偏倚风险评估工具,对SYRCLE工具发表前后的动物实验方法质量进行全面评估和对比,分析发现差距和存在的问题,以促进动物实验结果的转化和利用。方法计算机检索Web of Science和中国科学引文数据库(CSCD),检索时限为2011年起至2018年10月止。由4名研究者独立筛选文献、提取资料,并采用SYRCLE工具评价纳入研究的偏倚风险。计数资料采用例数和百分比(%),计量资料采用中位数和四分位数间距进行统计描述,组间比较采用卡方检验。采用SPSS 23.0进行统计分析。结果最终纳入5987篇动物实验。评估结果显示:SYRCLE工具涉及的10个条目中,60%(6/10)的条目的"低风险"符合率低于30%,其中66.67%(4/6)的条目的"低风险"符合率不足10%。尽管2014年后发表的动物实验在"分配序列、基线特征、动物安置随机化、不完整数据报告和其他偏倚"方面得到改善和提高(P<0.001),但在"隐蔽分组"方面差异无统计学意义(P=1),在"随机性结果评估和盲法"方面"低风险"符合率甚至低于以往的研究(P<0.001)。结论自SYRCLE偏倚风险评估工具发表以来,国内外动物实验研究在方法学方面仍存在一些问题,其中在随机性结果评估和盲法的应用和实施方面的问题尤为突出。建议采取针对性措施进一步推广与使用SYRCLE动物实验偏倚风险评估工具,以有效指导动物实验的研究设计与实施,提高动物实验的质量并促进结果的转化和利用。

Objective To comprehensively assess and compare the methodological quality of animal experiments published before and after the publication of SYRCLE’s risk of bias tool,find gaps and problems,thereby promoting the conversion and utilization of the results of animal experiments.Methods The Chinese Scientific Citation Database(CSCD)and Web of Science were searched from 2011 to October 2018.Document screening and data extraction were performed independently by four researchers.The methodological quality of the included studies was assessed using the SYRCLE’s risk of bias tool.Statistical analysis was performed by SPSS23.Results A total of 5987 animal studies were included.Methodological quality assessment showed that 60%(6/10)of the items were rated as"low risk"in less than 30%of the included studies,of which 66.67%(4/6)were rated as"low risk"in less than 10%of the included studies.Although the quality of animal experiments published after 2014 was improved in"sequence generation,baseline characteristics,random housing,incomplete data reporting and other bias"(P<0.001),there was no significant difference in"allocation concealment"(P=1),and the"low risk"coincidence rates in"random outcome assessment and blind method"were even lower than that in previous studies(P<0.001).Conclusion The published animal experiments still have some methodological limitations,especially in random outcome assessment and blind method.Therefore,we suggest that it is necessary to take targeted measures to promote and popularize SYRCLE’s risk of bias tool,thereby improving the design and implementation of animal experiments.