Bayesianadjustmentforover-estimationandunder-estimationofgastriccancerincidenceacrossIranianprovinces

摘要： AIMTocorrectthemisclassificationinregisteredgastriccancerincidenceacrossIranianprovincesincancerregistrydata.METHODSGastriccancerdataisextractedfromIranianannualofnationalcancerregistrationreport2008.ABayesianmethodwithbetapriorisimplementedtoestimatetherateofmisclassificationinregisteringpatientspermanentresidenceinneighboringprovince.Eachtimetwoneighboringprovinceswithlowerandhigherthan100%expectedcoverageofcancercasesareselectedtobeenteredinthemodel.Theexpectedcoverageofcancerouspatientisreportedbymedicaluniversityofeachprovince.Itisassumedthatsomecancercasesfromaprovincewithalowerthan100%expectedcoverageareregisteredintheirneighboringprovincewithmorethan100%expectedcoverage.RESULTSTheconditionwastruefor21provincesfromatotalof30provincesofIran.Itwasestimatedthat43%ofgastriccancercasesofNorthandSouthKhorasanprovincesinnorth-eastofIranwasregisteredinRazaviKhorasanastheneighboringfacilitateprovince;also72%misclassificationwasestimatedbetweenSistanandbalochestanprovinceandRazaviKhorasan.Themisclassificationratewasestimatedtobe36%betweenWestAzerbaijanprovinceandEastAzerbaijanprovince,21%betweenArdebilprovinceandEastAzerbaijan,63%betweenHormozganprovinceandFarsprovince,8%betweenChaharmahalandbakhtyariprovinceandIsfahanprovince,8%betweenKogiloyeandboyerahmadprovinceandIsfahan,43%GolestanprovinceandMazandaranprovince,54%betweenBushehrprovinceandKhozestanprovince,26%betweenIlamprovinceandKhuzestanprovince,32%betweenQazvinprovinceandTehranprovince(capitalofIran),43%betweenMarkaziprovinceandTehran,and37%betweenQomprovinceandTehran.CONCLUSIONPolicymakersshouldconsidertheregionalmisclassificationinthetimeofprogrammingforcancercontrol,preventionandresourceallocation. ... （共7頁）