Meta-analysis of a study on the efficacy of brucella treatments and publication bias

Meta-Analysis a method of combining the results of multiple studies which are made independently of each other on a specific isue and making statistical analysis of obtained research findings. The meta-analysis gives different results according to different approoches for disappearance of such problems appropriate research methods should be considered and appropriate parameters estimation method should be selected. Also to estimate how many research should be in the meta-analysis results. RCT?s are studies which are designed in developed countries and meta-analysis studies are produced through the diseases which are frequent in these countries. For this reason the meta-analysis studies about the frequent diseases in devoloping countries, Turkey and mediterrinean countries is one of this.The purpose of this study, understanding the basic principles underlying the meta-analysis studies, discussing which parameters to use and consider what while executing a meta-analysis study and evaluating the published meta-analysis reports, revealing the advantages and deficiencies of the proposed models (Fixed Effect or Raondom Effect) in meta-analysis, making the publication bias assesment practices by looking the distorting effect of missing data in meta-analysis and contributing meta-analysis studies which are done lw in our country.In this research, a meta-analysis is done by different researchers about effective antibiotic combinations in brucella treatment in different times between 1980- 2010 across the globe. In determination of Brucella treatment in adults, randomized controlled studies are considered; 104 randomized controlled studies are found about this subject and 32 out of 104 are included to the research.In this research, Mantel-Haenszel fixed effect model, Dersimonian and Laird random effect model and Peto method, meta analysis? models were used for testing homogeneity and influence quantity of antibiotic methods used in brucella treatment is acquired by imputed value meta analysis methods.Tetrasiklin?Stereptomisin and Tetrasiklin-Rifampisin treatments were also compared in relapses. Comparison was made with 943 patients and 12 trials. In Tetrasiklin- Rifampisin treatment relapse is significantly high. Mantel-Haenszel fixed effect model?s Odds Ratio is 2,67 (%95 GA:1,63-4,37) and in general comparison significant heterogeneity couldn?t be found in comparison made for general failure (p=0,914, I2=% 0). DerSimonian-Laird random effect model?s Odds Ratio is 2,53 (%95 GA:1,54-4,20) and in general comparison significant heterogeneity couldn?t be found (p=0,916, I2=% 0). Peto method?s Odds Ratio is 2,71 (%95 GA:1,70-4,32) and significant heterogeneity couldn?t be found also for general failure (p=0,791, I2= % 0).According to the results, all methods that used to ensure more reliable interpreation (Mantel-Haenszel fixed effect model and Peto method) sign that number of studies should be more then thirty in the comparison of general failure of the Tetrasiklin-Stereptomisin and Tetrasiklin-Rifampisin treatment methods.Analyses which is performed with only one method might not have always an accurate opinion about the prediction of the parameters. When the meta analysis? studies are done, prediction of parameters sould be obtained with all methods and the final decision must be given by considering all these results.Key Words: Treatment of Brucella - Meta Analysis - Publication Bias