CHAPTER6EquivalenceTests6.1IntroductionSometimesthegoalofaclinicaltrialistoestablishequivalencebetweentwotreatmentsratherthanthesuperiorityinefficacyofoneovertheother.Forexample,ifanewtherapyislesstoxicorlessexpensivethanthestandard,itmaynotbenecessarytoproveitisalsomoreeffective;instead,itcansufficetodemonstratethatitisequallyeffectiveorthatitislesseffectivethanthestandardbyatmostasmallamount.Dunnett&Gent(1977)giveanexampleofatrialinwhichanobjectivewastodeterminewhethercertainaspectsofhealthcarecouldbedeliveredequallywellbyasysteminvolvingatriageprocessbynurse-practitionersasbyconventionalprimaryphysiciancaremethods.Otherexamplesinclude:lumpectomyversusradicalmastectomyforbreastcancertreatment,alocalrice-basedbeverageversusWHOoralrehydrationsolutionfordiarrheaininfantsinMexico,andlow-doseAZTtreatmentversusstandarddoseduringpregnancyanddeliverytopreventmother-to-childHIVtransmissioninadevelopingcountry.Wecallthese“one-sidedequivalence”problems.A“two-sidedequivalence”problemcanariseintheareaofbioequivalencetesting.Here,apharmaceuticalmanufacturerhopestodemonstratethatanewpreparationofadrughasthesamebioavailabilitypropertiesasastandard,withinasmalltolerancelimit,asasteptowardprovingthatthenewandstandardpreparationshaveequaltherapeuticeffects.Demonstrating“bioequivalence”inthiswaycangreatlyreducetheamountofexperimentationrequiredforapprovalofanewdrug.Chinchilli(1996)classifiesequivalenceproblemsintothefollowingcategories:Populationequivalence:theresponsesforbothtreatmentshavethesameprobabilitydistribution;Averageequivalence:theresponseshavethesamemean;Individualequivalence:theresponsesare“approximately”thesameforalargeproportionofsubjectswhentheyreceiveeithertreatment.InSections6.2and6.3weshallbeconcernedprincipallywithaveragebioequivalence;weconsidertestsofindividualequivalenceinSection6.4.Ofcourse,toconcludeequivalence,itisnotsufficientthatthedatafailtorejectanullhypothesisofequality.Thismightsimplybeduetoalackofpowerofthestudy,andapracticallysignificantdi...
