《中国循环杂志》刊出论文的统计学方法评价

Evaluation of Statistical Methods for Articles Published in《Chinese Circulation Journal》

目的:评价《中国循环杂志》刊出论文的统计学方法的使用情况及存在问题。方法:对《中国循环杂志》2010年、2011年和2015年发表的论著类文章中统计学方法使用情况进行收集、分析和评价,发现存在的问题,并在不同年份间进行比较。结果:共404篇文章纳入评价,其中366篇使用了统计检验方法,32篇仅进行了描述性统计,另有6篇论著为Meta分析。使用频率较高的统计检验方法依次为:t检验(57.4%)、卡方检验(56.7%)、方差分析(35.2%)和回归分析(23.0%)。方法学部分对统计学方法描述存在的主要问题有:描述的方法与实际使用的方法不匹配、描述存在欠缺或错误。其中描述存在欠缺或错误的比例在2010年和2011年为36.6%(71/194),2015年为26.2%(55/210),差异有统计学意义(P=0.02)。统计检验方法使用错误主要为忽略研究设计错用统计方法、忽略统计方法使用条件等,统计方法使用错误的比例在2010年和2011年为19.1%(33/173),2015年为19.7%(38/193),两年份间无明显差异(P=0.88)。结论:2015年刊出的论文与2010年和2011年相比,统计学方法正确使用情况有所改善,但仍存在不同程度的问题,应进一步加强论文的统计学审稿工作。

Objective： To evaluate the applications and problems of statistical methods for articles published in《Chinese Circulation Journal》. Methods： We reviewed the statistical methods for original articles published in《Chinese Circulation Journal》in 2010, 2011 and 2015; general status with problems was analyzed and the situations were compared among different years. Results： A total of 404 articles were enrolled, statistical hypothesis testing was performed in 366 of them, 32 had descriptive statistics only and 6 articles were meta-analysis. The most frequently used statistical methods were student＆#39;st-test （57.4%）, followed by chi-squared analysis （56.7%）, analysis of variance （35.2%） and regression analysis （23.0%）. The main statistical problems in ＆quot;Methods＆quot; included in described method un-matched to really used method and insufifcient or wrong description. The proportion of insufifcient or wrong description in 2010 and 2011 was 36.6% （71/194）, while in 2015 was 26.2% （55/210）,P=0.02. The major wrong application of statistical approach was ignoring design with improper method and using unsatisifed precondition for speciifc analysis. The proportion of improper statistical method application in 2010 and 2011 was 19.1% （33/173）, in 2015 was 19.7% （38/193）,P=0.88. Conclusion： The quality of statistics was improved in 2015 than 2010 and 2011 for articles published in《Chinese Circulation Journal》; while there were still problems at certain degree, enhanced statistical review should be conducted in medical journals.