临床医学论文中的统计问题

发布时间：2018-02-14 12:21

  本文关键词： 科研设计 可比性 两类错误 置信限 混杂 统计软件　出处：《数理统计与管理》1998年01期 　论文类型：期刊论文

【摘要】：本文对中华医学会主办的１４种临床医学和基础医学杂志发表的１５５篇论著，从医学统计的角度，，按２２项指标，进行了一次调查研究。结果表明，与１９８０年中华医学会召开医学科研设计座谈会以前的情况相比，论文质量有了明显提高，但仍存在不少不可忽视的问题。常见的问题有：试验设计不良，缺少防止偏性的有力措施，组间可比性差，样本含量不足；对两类错误的控制、置信限的应用和排除混杂因子的影响重视不够；统计软件、统计图和多元分析方法的利用率还不够高。针对存在问题提出了一些见解和建议。
[Abstract]:In this paper, a survey of 155 articles published in 14 clinical and basic medical journals sponsored by the Chinese Medical Association has been carried out according to 22 indexes from the point of view of medical statistics. The results show that, Compared with the situation before the medical scientific research design symposium was held by the Chinese Medical Association in 1980, the quality of the paper has been obviously improved, but there are still many problems that can not be ignored. Lack of effective measures to prevent bias, poor comparability among groups, insufficient sample size, insufficient attention to the control of two types of errors, the application of confidence limits and the impact of eliminating confounding factors; statistical software, The utilization rate of statistical charts and multivariate analysis methods is not high enough.
【作者单位】： 中日友好医院 中华医学会 中日友好医院
【分类号】：C931.1


本文编号：1510694