分析临床研究中等级资料统计存在的问题与对策  

Analysis problems with solutions for ranked data in clinical trial

作　　者：毕京峰[1] 陈素红[1] 屠舒[1] 朱真珍[1] 魏振满[1] 

机构地区：[1]中国人民解放军第三〇二医院,北京100039

出　　处：《中国临床药理学杂志》2012年第10期776-778,共3页The Chinese Journal of Clinical Pharmacology

基　　金：国家十一五科技重大专项课题基金资助项目(2009ZX10005-017);中国人民解放军第三〇二医院院内课题基金资助项目

摘　　要：目的探讨临床研究中等级资料的统计分析方法。方法基于临床研究目的和统计分析原理,应用模拟数据进行分析。结果部分研究数据分别应用等级资料的秩和检验及有效率的χ2检验,会得出相反的统计结果;部分具有前后测量的等级资料,仅对治疗后数据应用Ridit分析或Wilcoxon秩和检验进行统计分析,忽视了治疗前等级情况对统计分析结果的影响。结论应根据研究目的,确定应用等级资料的统计方法,对有前后测量的等级资料,要用等级资料前后测量的统计分析方法。Objective To discuss the statistical analysis method of ranked data in clinical research. Methods Been analysised with simula- ted data based on clinical research purposes and statistical analysis prin- ciple. Results The statistical analysis results may be different if Wilcox- on rank sum test and Chi - square test are used respectively of some ranked data. Of some repeated ranked data, the baseline before treatment was neglected if only analysis the data after treatment with Ridit test or Wilcoxon rank sum test. Conclusion It is recommended that based on the purpose of a research project, the statistical analysis approach： appli- cable approach for ranked data or based on Pearson~ chi - square test. If ranked data consists of pre - and post -treatment measurement informa- tion, statistical analysis of post - treatment data with Ridit analysis or Wilcoxon rank - sum test may be inaccurate. It is advised to adopt an ap- plicable statistical approach to analyze both pre - and post - treatment measurement information.

关 键 词：等级资料 统计分析 

分 类 号：R969.4[医药卫生—药理学] R195.1[医药卫生—药学]