Analysis Of Statistical Methods Used In The Manuscripts Of One Medical Journal

To survey and count202manuscripts from Aug,2008to Dec,2008about their applications of statistical methods in which contents involved clinical research, zoopery and preclinical medicine. The indicators include the following contents:first, whether the the research methods, baseline information and the data descriptions used in the manuscripts were correct; second, the descriptions and despressions of statictical methods, the data processing tools and whether the statistics despressed P-value; third, whether the statistical methods used in the articles were correct and types of data desining and the statistics method which should be used. We only count the statistics expressions of the manuscripts except the baseline information. To research the problems of authors about statistical methods,their research level, and the requirement to medical editors, We aim at discussing how to improve the medical research level of medical researchers and the research quality of medical editors, as well as the CME ways of medical statistics.The202authors spread over23provinces,39clinical and preclinical rooms. Only7articles refered to research method simply but the calculation of sample content, they didn’t pay enough attention to control group, lacked adequate awareness about the balance and comparability of baseline information, and the study factors were uncertainty. Using mean and standard deviation to describe abnormal distribution is the principle problem of data description. The articles mentioned statistial method for315times including14kinds of methods, there were180correct kinds and135wrong among those methods. There were32times of using statistical methods but not mentioned in the articles.There were194times cited measurement datas as indicator in the results of202articles, among which the information of compare means in two groups of independent samples were indicated for63times at most, next were repeated measurement and design datas for36times,unbalance composite designition data for33times, monothetic and multilevel datas for25times, relevance between dichotomous variable for22times, and others for15times. Cited enumeration data as indicator for92times, among which the comparaton between two-sample and multi-sample for69times, including some repeated datas, and others for23times. Cited ordinal data as indicator for33times, among which the the comparation of therapeutic effect between groups for25times, others for8times. The percent of articles about149pieces that described statstical methods elaborately was73.8%, but61.4%about124pieces of them didn’t put forward statistics datas.86.1%of202articles put forward the range of P-value, but only10.4%of them included precise P-value. It is nonconformance with the condition that77.2%(156/202) using statstical softwares to make statistical treatment.The error rate of common statistical methods is high, however, on the contrary, the accuracy rate of varray multivariate statistical analysis is higher. The main reason of using wrong statistial methods is that the authors used various methods uncertainly even chose simple methods, neglected application conditions of each methods and research purpose as well as the study factor. Such conditions show that the research and statistics level of clinical research specialist staff must be improved. The defect of research design and statistical skills set back the in-depth exploration of clinics treatment research of relevance staffs and restrict their ability of refering to medical literature, the level of medical journals becomes lower in turn.Therefor, we should strengthen the continuable medical statistics education to clinical research staffs and medical editors, which contents emphasize particularly on statistical methods and application conditions, and teach them how to choose proper description and deduction methods,or to draw a clear distinction between statistical results and professional conclusion, and how to express the results succinctly in research literature. Because of the important effect of medical editors, the medical editorial office could organize statistics CME to authors, such the research level of authors and editors will be all improved through thesis writing.