Investigating The Publication Bias Of The Meta Analysis About Medicine In Chinese Science Citation Database

Objective: Publication bias(PB)is an important factor of affecting the quality of meta-analysis,and whether the publication bias diagnosed can be used as one of the criteria to evaluate the quality of articles.The purpose of this study is to construct an objective and impartial PB evaluation method by third party and investigate the diagnosis of PB of the Meta analysis in Chinese Science Citation Database(CSCD),compare the selection of diagnosis method and diagnosis results of third party and original authors,evaluate and monitor PB of the articles by third party,it provides reference for scientific researchers to choose the appropriate publication bias diagnosis method.Methods: Set the five parameters,the kind of effect size,the number of study,the values of effect size,the mean and standard deviation of sample size of studies,and then do the data simulation,create meta-analysis simulation data with PB and simulation data without PB,calculate the sensitivity and the specificity of eight candidate methods in different parameter conditions.By calculating the weighted average of the sensitivity and the specificity,find out the best methods among the eight methods to detect the publication bias in permutation of the four parameters,established the evaluation method of publication bias of meta-analysis.Exploring the published medicine Meta-analysis literatures(as of August 2017)in Chinese Science Citation Database(CSCD),investigating the diagnosis of publication bias(PB)of the medicine meta-analysis,using R 3.3.1 to calculate the results of the publication bias diagnosis when different diagnostic methods applied,based on the evaluation method of publication bias in meta-analysis established by the third party,in combination with the four parameter values of the meta-analysis(high,medium and low)to select the best diagnosis method,using SAS 9.3 to do Comparative analysis of diagnosis method selection and diagnosis results of third party and original authors.Results:Constructing a PB evaluation method by third party,according to the application tool table of candidate publication bias diagnostic method and permutation of the four parameters in every meta-analysis,selecting the best method from the eight candidate methods.Only 10% of the Meta-analysis in CSCD has carried on the diagnosis of publication bias,which is far lower than the articles in the Science Citation Index expanded,and most of the authors chose egger or begg method as a publication bias diagnosis method which is different from the third party,the rate of the definite publication bias of the authors(0 percent)is lower than that of a third party(20 percent)which use the evaluation method.Conclusions: Constructing a PB evaluation method by third party,can be used in diagnosis of published meta-analysis paper.The diagnosed rate of publication bias of Meta-analysis in CSCD was relative low,perhaps associated with the impact factor of articles' published journals.The diagnosed rate of publication bias of Meta-analysis in CSCD was lower than that of a third party.Suggested that reviewers should strengthen verification in this aspect and strengthen the train of evidence-based medicine methods for related professionals.