A Review of the Statistical Analysis used in Clinical Articles Published on Journal of Korean Neurosurgical Society

Kang, Weechang

Abstract

Statistical analyses used in clinical articles published on the Journal of Korean Neurosurgical Society were identified and appropriateness of statistical aspects in reporting results was assessed. Forty seven clinical articles were selected in this study, which were published from February, 2005 to February, 2006 on the journal. The frequency of statistical analysis was as follows: descriptive statistics only 24 (51.1%), one type of statistical method 10 (21.3%), two or more methods 13 (27.6%). An assessment of statistical aspects was performed in 24 clinical articles reporting inferential statistics. Ten articles (41.7%) did not adequately describe or reference all statistical methods used. There were six articles (25.0%) not reporting the confidence level used as the critical criteria of the statistical significance. In thirteen articles (54.2%) it seems more appropriate to implement multivariate analyses in addition to univariate analyses. We recommend that the journal readers should concentrate on improving their knowledge of basic statistics and statistical review for manuscripts submitted should be sought from professionals in the fields of biostatistics and epidemiology.