Meta-analysis Method And Its Application Of The Decision Of Epidemiology

Meta-analysis is a method that combines summary quantitative information from from related problem but independent studies.By considering the differences of within-study and between-study and the publication bias of the studies,this method obtains a summary review which solve the problem of the disadvantages of the single study,partic-ularly with regard to discrepancies in the ethnicity,geographical region and the number of the samples.it's conclusions that the results of Meta-analysis play an important role for many fields of the science research.In this thesis,we study the following two problems:(1).The method of bivariate Meta-analysis based on the difference mean;(2).Application of Meta-analysis on the influence of IL10 Gene polymorphisms on non-Hodgkin lymphoma and chronic hepatitis C to PEG-interferon/ribavirin therapy.The conclusions of these problems are the following detailed results:Firstly,we establish the fixed-effect model of bivariate Meta-analysis based on the difference mean.Under the assumption of the multilevel normal distribution of the differ-ence mean,we obtain the maximum likelihood estimator ?ML of the difference mean by the method of maximum likelihood estimate.We conclude that the maximum likelihood estimator ?ml of the difference mean is unbiased for the difference mean effect ?.In addition,we get the corresponding covariance matrix and the(1-?)100%confidence interval for the difference mean effect ?.Under the framework of this model,we get the following unknown variance models by two steps:Step I,we obtain the maximum like-lihood estimator of the difference mean effect and unknown variances;Step II,we find the combining overall difference mean effect Doverall by the fixed-effect model of bivariate Meta-analysis based on the difference mean.Moreover,we obtain the corresponding co-variance matrix and the(1-?)100%confidence interval for the overall difference mean effect Daverall.Secondly,we establish the random-effect model of bivariate Meta-analysis based on the difference mean.Under the assumption of the multilevel normal distribution of the difference mean,we obtain the maximum likelihood estimator ?ML of the difference mean by the method of maximum likelihood estimate.We conclude that the maximum likelihood estimator ?ML of the difference mean is unbiased for the difference mean effect?.In addition,we get the corresponding covariance matrix and the(1-?)100%confidence interval for the difference mean effect ?.Thirdly,we subsequently performed a meta-analysis of all eligible case-control studies to determine a more precise estimation of the association of IL-10-3575T/A,-1082A/G,-819C/T,and-592C/A polymorphisms and their haplotypes with the NHL susceptibility.Based on the results of the meta-analysis,we concluded that:(1)IL-10-3575T/A and-1082A/G polymorphisms were associated with altered NHL susceptibility,especially for Caucasians and B cell lymphoma.(2)The A-G-C-C haplotype was associated with a significantly increased risk of NHL and DLBCL,and T-A-C-C haplotype was associated with a significantly decreased risk of DLBCL.Finally,we subsequently performed a meta-analysis of all eligible case-control studies to determine a more precise estimation of the effects of IL-10-1082A/G,-819C/T,and-592C/A polymorphisms and their haplotypes on sustained virological response(SVR)in chronic hepatitis C(CHC)patients receiving pegylated interferon alpha(PEG-IFN-a)plus ribavirin.Based on the results of the meta-analysis,we concluded that:(1)IL10-1082GG genotype and-1082G allele were associated with decreased SVR rate in CHC patients,especially for the Egyptian and HCV-4 genotype.(2)IL10-819T allele was more likely to get SVR in the Caucasian population.