Statistical Methods in the Articles in the Korean Journal of Anesthesiology Published from 1994 to 1998

Ahn, Wonsik

Abstract

BACKGROUND: The Korean Journal of Anesthesiology (KJA) was first published in 1968 containing only 16 articles. In 1998, the number is 291. However, the quantitative growth does not mean a qualitative growth. There are many aspects to improving quality. One of them is statistical accuracy. I have examined the statistical methods in our field and checked the accuracy of the methods. Then, I compared the results with the methods in the 1980s and examined what kinds of changes exist. METHODS: I reviewed all the articles except case reports and review articles in KJA published from 1994 to 1998. I focused on the methods of inferential statistics because those kinds of statistics were usually mentioned in the articles. It is based on the mentioned technique in the article to decide which inferential statistics are used, even though sometimes they are not accurate. I adopted the similar statistical error criteria selected by Ko. RESULTS: Basic statistical error, for example no statistics used even though statistical methods are needed, were dramatically reduced compared to the 1980s. It is increased to use the mean comparison methods correctly, but, some statistical methods are still misused frequently, for example Chi-square test, nonparametric analysis, multiple comparison methods and improperly adopted methods based on the variable scale. CONCLUSION: Generally, based on my criteria statistical errors are reduced from about 75% in the 1980s to about 60% in the 1990s.