An Assessment of Statistical Methods in the Journal of Korean Child Neurology Society, 1993-2003

Kim, Joon Sik; Kwon, Tae Chan; Lee, Choong Won

Abstract

PURPOSE: A clinical trial cannot be adequately interpreted without information about the methods used in the design of the study and the analysis of the results. We would like to evaluate trends in statistical methods and describe the frequency with which various statistical techniques are reported in the Journal of Korean Child Neurology Society.
METHODS
We reviewed 288 original articles published in the Journal of Korean Child Neurology Society from 1993 to 2003 to assess the statistical methods.
RESULTS
The number of cross-sectional study was 232(80.6%) articles and that of animal study was 45(15.6%) articles but Cohort study was only eleven(3.8%) articles. One hundred twenty seven(44.1%) articles used no statistical methods or descriptive statistics only, frequency of which decreased yearly and 84 articles(29.2%) used T-test, frequency of which increased yearly. Seventy two(25%) articles used contigency tables and twenty three(8.0%) articles used ANOVA. Orphan P where no statistical methods had been specified and only the P value given was presented in 19 articles(6.6%) which decreased yearly.
CONCLUSION
These results suggest that medical articles published in the Journal of Korean Child Neurology Society, 1993-2003, were short of their expected quality and the validity of statistical methods used appears to be seriously compromised in this period and has much to be done to improved the current situation. It is concluded that a basic training in biostatistical methods, more consultation of medical investigators with statistician or other experts, careful review by someone in biostatistics or research design before accepting a manuscript are needed.