高影响力期刊2014～2016年动物实验研究的方法学质量评价

The Evaluation of methodological quality of animal studies in high impact journals from 2014 to 2016

目的评价国内外高影响力期刊已发表动物实验研究的方法学质量，为提高动物实验的方法学质量和生产高质量的临床前研究提供参考。方法计算机检索中国科学引文数据库（CSCD）和Web of Science数据库，搜集所有涉及治疗干预措施的动物实验，检索时限为2014年至2016年8月。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方法学质量。结果最终纳入1999个动物实验。超过90％的研究被引频次≤5次，52．53％的研究被引频次为0。SYRCLE工具评价结果显示：54．55％的亚条目的“低风险”的符合率不足30％，其中84．62％的亚条目的“低风险”的符合率低于10％。结论国内外动物实验研究在方法学设计方面均存在一定问题，国内动物实验研究在实施偏倚、测量偏倚、失访偏倚等方面的问题尤为突出，其“低风险”符合率远低于国外研究。建议采取针对性措施推广与普及SYRCLE动物实验偏倚风险评价工具，以有效指导动物实验的开展，提高动物实验的设计和实施质量。

Objective To evaluate the methodological quality of animal experiments published in high impact journals, in order to provide references for improving the quality of animal experiments. Methods CSCD and Web of Science databases were electronically searched to collect intervening primordial animal experiments from 2014 to August, 2016. Four reviewers independently screened literatures, extracted data and assessed the methodological quality of included studies by using SYRCLE tool. Results A total of 1 999 animal experiments were included. The cited frequency of more than 90% studies were ≤ 5 times, and of which 52.53% studies were zero. The results of SYRCLE evaluation showed that 54.55% of sub items rated as ＂low risk＂ were less than 30%. And 84.62% of them were less than 10%. Conclusion There are defect in methodological quality of animal experiments either domestic or abroad. The problems of domestic researches in implementation bias, measurement bias and loss of access bias are particularly obvious. The coincidence rates of＂low risk＂ are much lower than those of abroad studies. Therefore, we suggest that it is necessary to take specific measures to popularize SYRCLE tool to effectively guide the development of animal experiments and improve the design and implementation of animal experiments.