An Assessment of Methodological and Ststistical Validity of Medical Articles Published in Korea, From 1980, To 1989

Lee, Hyeong Ki; Ahn, Yoon Ok

Abstract

The methodological and statistical validity of 382 original articles published in the Journal of the Korean Medical Association, from January 1980 to December 1989 was reviewed by the author-devised c heck list consisting of 21 items (14 items for methodological validity and 7 items for methodology, and of 297 articles using statistical analyses a total of 290 articles (97.6%) were found to contain at least one error in statistical methods used. The mean and standard deviation of 'validity score of one article', defined as the total number of valid items devided by the total number of applicable items and then multiplied by 100, were 43.8 and 15.2, respectively. The distribution of validity score was as follows ; over 60 (57 articles, 14.9%), 30 to 59 (266 articles, 69.6%), and under 30 (59 articles, 15.5%). The proportion of articles, of which validity score was over 60, was significantly higher in descriptive study (19.4%) than in analytic study (8.4%, p = 0.003). Also the articles of over 60 in validity score were more frequent in survey (15.9%) than in experiment (8.2%), and in cross-sectional study (16.8%) than in longitudinal study (10.6%), but this finding was not statistically significant. The averaged validity score of two year period was highest in 1984-1955 (50.24), and lowest in 1986-1987 (38.85). There was no significant time trend of the averaged validity score over 10 years(p>0.1). These results suggest that medical articles published in Korea, 1980-1989, were short of their expected quality, and there have been no evidence of improvement with time. It is concluded that a basic training in biostatistical methods in the medical postgraduates and residencies, more consultation of medical investigators with statistician or other experts, and careful review by someone knowledgeable in biostatistics or research design before accepting a manuscript are needed. In addition, refutation should be allowed for the controversial point through the journal.