临床科研中常见的统计误用问题

• 作 者: 凌瑞珠
• 机 构: 北京医学院临床医学研究所医学统计室,
• 出 处: 中国卫生统计
• 发表日期: 1985年02期
• 关键词: 国内医学 英国医学杂志 文献评论 中毒型痢疾 人工冬眠 基本治愈率 文中 样本含量 科研设计 疾病诊断

在“怎样发现、纠正和防止医学文献中的错误”一文中有这样一段话,“生物医学文献评论者经常发现约半数文章所使用的统计方法是不正确的”。文中列举《英国医学杂志》1976年三个月内的文章中有42%至少有一处错误;《循环》1977年7～12月的原著中27%使用的统计方法不正确。本人曾系统阅读1978～1979年中华医学检验杂志,其中原著中50～70%存在统计误用,其它国内医学杂志的统计误用也多处于类似水平。

In the article “How to Find, Correct, and Prevent Mistakes in the Medical Literature”, “Biomedical literature reviewers often find that the statistical methods used in about half of the articles are incorrect.” In the article, 42% of the articles published in the British Medical Journal in the three months of 1976 contained at least one error; 27% of the original works in the “Circulation” from July to December 1977 were incorrect. I have systematically read the Chinese Medical Examination Magazine from 1978 to 1979, in which 50 to 70% of the original works were misused by statistics, and the statistical misuses of other domestic medical journals were also at similar levels.