CMH检验和Meta分析在临床试验中心效应分析的应用探讨_欧爱华.pdfVIP

# 50# Chinese J Ev idence-BasedM edicine, 2006, Vol. 6( 1) CMHMeta 欧爱华 老膺荣 何羿婷 何晓红 王宝娟 ( 510120) , CMH M eta CMH (P 01 05)M eta ( V2 = 121 1, P = 0112 ), , (Z = 3135, P 01 05) , , M eta, M eta, , CMH ; CMH ; M eta Application ofCMH Test andM eta-analysis inMulti-center Randomized Controlled Trials OU A-i hua, LAO Y ing- rong, H E Y -i ting, HE X iao-hong, W A NG B ao- juan. Guangdong ProvincialH osp i- tal of Trad itional ChineseM edicine, Guangzhou 510120, China Abstract Objective To seek the statistical solution in the comparison of different effects from m ulti-center ran- dom ized controlled trials ( RCTs). M ethods The data collected from a mu lt-i center RCT were used as the examp les and processed by CMH test and m eta-analysis. Results The result of CMH test ind icated that the sign ificant difference of the effect values existed am ong centers (P 01 05). Wh ilem eta-analysis showed no sign ificant difference (P 01 05) by het- erogeneity test. H owever, w hen using fixed effect m ode,l inter-group sign ificant difference ofm erged effect valuesw as ob- served (P 01 05). Conclusions In the clin ical research based on them ethod ofmu lt-i center RCT, meta-analysis can be app lied if the difference of inclination of the inter-group therapeu tic effect is found am ong d ifferen t centers. The proper m athem aticalm odel shou ld be selected based on the result of heterogeneity test tom erge and compare the effect values. The conclus ions shou ld be draw n from the results of both m eta-analysis and CMH test. eywords Mu lt-i center random ized con trolled trials; CMH