986DSFREDRICKSONwhethersufficientnumberswilltakepartandstaylongenoughtoseethetrialthrough4Thefourthisinterpretationthepostgamecritiquethatdetermineswhatwehavelearnedandwhetherweshallbeginallover ID: 955592
Download Pdf The PPT/PDF document "985THEFIELDTRIALSOMETHOUGHTSONTHEINDISPE..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
985THEFIELDTRIAL:SOMETHOUGHTSONTHEINDISPENSABLEORDEAL*DONALDS.FREDRICKSONDirectorNationalHeartInstituteBethesda,Md.THEfirstrecordedfieldtrialtookplacemanyyearsagointheGardenofEden.Theexperimentalprotocolhadreceivedinade-quatepriorconsideration,thepopulationsamplewastoosmall,thestudyconsumedtoolargeafractionofthegrossnationalproductatthetime,andtheresults,whileconsideredbysometobeunequivocal,havegenerallybeenfelttohavebeenoverinterpreted.Andthereyouhaveasummaryofmostofthecriesraisedagainstfieldtrialsinbiologyfromthebeginningtothepresentday.Sinceeverypracticingphysiciansuffershistrialinthefieldnearlyeveryday,IhadbetterdefinewhatIpresentlyhaveinmindbytheterm"fieldtrial."Iamreferringtotheexperimentalapplicationofaninterventiontodetermineifitqualifiesasausefuladditiontothestandardpracticeofmedicineorpublichealth.Iamtalkingfurtheraboutlargetrials,forintheareaofcardiovasculardiseases,towhichmycommentswillberestricted,theuncertaintiesofourendpointsseemtomakeitderigueurtostudymanysubjects.Theingredientsforasuccessfulfieldtrialwillbevieweddifferentlybyexpertsofdifferentpersuasions.Frommyviewpointtheyareasfollows:I)Thefirstisproperdefinitionoftheproblem.Thishangsuponthepresenceofadequatecollectivedoubtontheonehand,andcollectiveagreementontheother.2)Thesecondingredientisstudydesign.Itsmanyfacetsshouldincludeanelementofcost-benefitanalysis.3)Athirdisparticipation,thecollectionoffactorsthatdetermine*PresentedaspartofaConferenceonCoronaryHeartDisease:PreventiveandTherapeuticAspectssponsoredbytheNewYorkHeartAssociation,NewYork,N.Y.,heldatTheWaldorf-Astoria,NewYork,N.Y.,January23,1968.Vol.44,No.8,August1968 986D.S.FREDRICKSONwhethersufficientnumberswilltakepartandstaylongenoughtoseethetrialthrough.4)Thefourthisinterpretation,thepostgamecritiquethatdeter-mineswhatwehavelearnedandwhetherweshallbeginalloveragain.Thislastphrasehasafacetiousringwhichisdeplorable.Yetitisagoodplacetobegin,forIbelieveitisnecessaryforustobecriticalofhowwellwehavedoneinthepast.Forexample,anumberoffieldtrialsinourdaythathavedealtwiththevalueofanticoagulantsorthee
ffectofvariousdrugsanddietsuponatherosclerosiscanbesaidkindlytohave"endedinequivocation."Theyinvitedthementalimageofawhite-coatedfigureendlesslyandunsuccessfullypursuingTruthacrosstheElysianfields.Itisnotthatwemusthavealwaysapositiveresultorthatweabhorthethoughtofanegative.Itisthedrawingofneitherthatissounsettling.Whatistheproblem?Ismantoocomplexforustoreducetomanipulationofsinglevariables?Shallweabandonfieldtrialsandrelyonintuitionwhenlaboratoryanimalswillnotdo?Theanswersareobvious.Maniscomplicated,butheisalsooftenuniqueasatestsystem.Fieldtrialsareanindispensableordeal.Theproblemistoconductbetterfieldtrials.Wecandothis,andIthinkwearedoingit.Wehavelearnedfromtrampingthroughthefield.Especially,Ishouldsay,sincebeingjoinedbythebiometricianssentoutbysocietytotherescue.Wehavelearnedthecostofcom-promise,ofcuttingnumbersofsubjectstoocloseinmisleadingattemptsateconomy,ofneglectingrandomizationandthedouble-blind,whentheyarefeasible,becauseofinconvenienceortheinterpositionoffalseissues,offailingtostandardizemethodsor,worstofall,ofhavingmis-guidedfaithinendpointsthatperniciouslysoftenwithtime.Learningbytrialanderrorispartoftheexperimentalprocess.Onecannotexpecteachexperimenttobeasuccess.Butwhentheexperimentisafieldtrialthecostsofanequivocalanswerareonadifferentscalefromthatofthelaboratory.Letmetaketwoexamplesofcost.ThelargestandmostcomplexfieldtrialnowsupportedbytheNationalHeartInstituteistheCoronaryDrugProject.Theaimistodeterminewhetheroneoffourdrugswilldecreasethemortalityinmenwhohavealreadyhadonemyocardialinfarction.Thefullstudyasprojectedwillrequire8,5oomen,eachparticipatingforfiveyears.Thisisatremendousundertaking.Forsuchastudy,whenitisfullyBull.N.Y.Acad.Med. THEFIELDTRIAL987~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~operating,morethanfourmilliondollarsayeararerequired.Theinstitutealsohasbeforeitthereportofadistinguishedandhard-workingpanelofexpertswhoconcludedfromtheirDiet-HeartFeasibilityStudythatthedietcanbealteredsoastolowercholesterollevelsinfree-livingAmericans.Theyfurtherrecommendedstronglythatsuchalterationsbecarriedoutinapopul
ationlargeenoughandforasufficienttimetodetermineonceandforallwhetherthisdecreasesthemortalityandmorbidityfromcoronaryarterydisease.Ifthisisdoneonafree-livingpopulation,excludingthosewhohavealreadyhadamyocardialinfarction,suchastudywilldwarftheCoronaryDrugProject.Costsarebeinganalyzed.ItisconceivabletheymightexceedIOoreven20milliondollarsayearforuptoioyears.Unthinkable?No,eminentlythinkable.Theeconomiccostofprematurecoronaryarterydisease(deathsfrommyocardialinfarctionbeforeage65)easilyexceedsIbilliondollarsayear.Apositivestepthatreducesthistollwillbeproportionatelyworthitincost.Ifitseemstoyouthatthenumberandgrandeurofproposedfieldtrialsisontheincrease,Ibelieveyouarecorrect.Andtheratewillcontinuetogrow,forseveraldecadesofintensifiedresearchactivitiesareinevitablycreatingmoreopportunitiesandmorepressuresforexploitationofbasicdiscoveries.Thenextdecadewillbringagreatmanymoresuchopportunities.Choosingamongtheincreasingoptionswillalsoposethemostdif-ficulttaskforthecommunityofexpertsandforsocietyatlarge.Societywillmakethefinaldecisions,foritbearsthecostsaswellastherewards.Wheretheoptionsinvolvecardiovasculardiseases,itisourjob,yoursandmine,tobeexceedinglythoughtfulandwiseinbothdecisionsandactionsappropriatetoourrolesasexperts.Societyhasreasontorelyuponus.Wehaveeverydesiretorespond.Indoingso,however,weareduty-boundtomovecarefullyandmountwithskillbarriersthathaveunhorsedourpredecessors.DEFINITIONOFTHEPROBLEMClearlythemostimportantfirststepinscientificinvestigationsistoask,"Whatisthequestion?"Inthefield,asinthelaboratory,thebettercircumscribedisthequestion,themorelikelyisonetoobtainanun-equivocalanswer.Thetendencytocomplicatethearrayofcanvasflowninproportiontothelengthofthevoyagehassentmanyashiptodis-Vol.44,No.8,August1968987THEFIELDTRIAL 988D.5.FREDRICKSONaster.Atinyboatsomismanagedsinkssilently.Whenamajorfieldtrialdismasts,itspillsavastcargoforalltosee;andthemournfulnoteoftheLutinebellclamorsonandoninourconsciences.Thereisanotherfeatureoffieldtrialsthatdiffersindegreefromlaboratoryexperiments.Oncethequestionhasbeenagreedupon,an
umberofparticipantsneedcontinuingassurancethataquestionexistsatall.Thisisunderstandable.ItisbecauseineachphysicianthereisatouchofFalstaffthatgrowsedgyattheconfrontationwithuncertainty.Gratifiedbythesuccessofamaneuverinasmallnumberofpatients,hemaymomentarilylosepatiencewithdoubtersandcryout:What,arethoumad?....Isnotthetruththetruth?*Buttheskeptic,whoispreeminentlythephysician,recovers;andweknowthatwedonotknow.Andthescientist,whoisalsothephysi-cian,knowsthatuncertainty,therestlessandunwantedguest,willhavetobeevictedintime.Theprocessofdispossessinguncertaintyispainfulandhoarywithconstraints.Itisofutmostimportancetonoteherethatafieldtrialisoftenarelativelylateexpressionofuncertainty.Itmayenterterrainalreadylitteredwithpreconceivedopinionandisallowedtoproceedonlybe-causeenoughcollectivedoubthasbeenaccumulatedtosecureitspas-sage.Theassemblingofcollectivedoubtandthecarewithwhichitisarticulateddeterminehowwellsocietycanaccepttheburdenofanexperimentamongitsmembersatlarge.Theproperfulfillmentofthistaskissurelyoneofthegreatestresponsibilitiesofmedicalscientistsandofthosewhomakerecommendationstosocietyaboutscientificpro-grams.Thecorollaryrequirement,ofcourse,isacarefullycollectedagree-mentthatthepossiblebenefitsofamaneuverbothcommenditfortrialandconsiderablyoutweighanyharmitmightdo.Forthispurposefairlyelaboratesystemshavecomeaboutinrecentyearsasconsciencehasbeencodifiedwithrespecttohumanrights.Theresponsibilitiesofthosewhoconductorprovidetheresourcesforinvestigationhavebeenmoreclearlydefined.Sufficientmechanismsforthejudgmentbypeersnowexist.Theyhavenotsolvedallforeseeableproblemsrelevanttofieldtrials,butatleastwearebetterabletograpplewithethicalissues;andweperceiveourlimitations.*HenryIV,part1,actII,scene4.Bull.N.Y.Acad.Med.988D.S.FREDRICKSON 989STUDYDESIGNIshouldhavelikedtoconcentrateontheingredientofstudydesign.However,itisthemarkofagoodamateurthatheknowshowtolosehisracquetinthepresenceofprofessionals;andIshallconfinemyselftoabriefrallywithoneortwooftheissues.Thefirstistheissueofthesizeofpopulationrequiredinatrial.Ihavealreadyme
ntionedthefrustratingobservationthatmanyfieldtrialscometoanendwiththeconclusionthatthenumbersweretoosmallforstatisticallyacceptableinterpretation.Thiscannotalwaysbeprevented.Butwheretheprobleminvolvedismortalityfromcoronaryarterydis-easetheexcuseforgrossuncertaintyhasdisappeared.Forthisweowerecognitiontoagroupofepidemiologicalstudiesthatthemselvesmatchinterventiontrialsintediousness,indemandforcarefulconstruction,andinindispensability.ButFramingham,Mass.,Tecumseh,Mich.,andtheirmanycongenersnowprovidepredictionsofincidenceandmor-talityandbasesforstratificationaccordingtoriskfactorsthathavemateriallyimprovedthescientificbaseforfield-trialdesign.Theanarchyofguessandintuitionhasgivenwaytoabenevolenttyrannyofstatisticians.Ihaverecentlystruckupapassingacquaintancewiththeirtwinwatchdogsofprobability,alphaandbeta,andIcom-mendallofyouwhohavenot,todothesame.Thesewatchdogshaveconsiderable,ifstilllimited,power.Inthesimplestterms(forsomeamplification,seeAppendix),whenwemakeacomparisonoftreatmentwithcontrol,alphaistheprobabilitythatthetreatmentisinferredtobesuperiortothecontrolwheninfactitisnot.TheCoronaryDrugProject,forexample,isdesignedinsuchawaythatsounfortunateaninferenceshouldnotoccurmoreoftenthanonetimeinahundred.Beta,whichpatrolstheotherborder,allowsustoassumethattherateofeventsinthetreatedgroupwill,infact,besmallerthantherateofeventsinthecontrols.It(beta)setsthelimitfortheprobabilitythat,intheactualtrial,thetreatmentratewillnotbesignificantlysmallerthanthecontrolrate.Itisthe"sensitivity"ofthetrial.IntheCoronaryDrugProjectbetais0.05.Allelsebeingequal,thechoiceofalphaandbetadeterminesthesamplesizeofafieldtrial.Otherfactorsmustbecounted,however,andeducatedguessesofthedegreeandtimingofthepresumedeffectoftherapymustbeadded.So,too,mustknowledgeofthepopulationbeVol.44,No.8,August1968989THEFIELDTRIAL 990D.S.FREDRICKSONadequatetopermitareasonableestimateoftherateatwhichsubjectswilldropoutofthestudy.Suchcomputationalproceduresarewellestablishedandthesettingoftheoddsinastudyinvolvingcoronaryarterydiseasenowapproachestheprecisionofawell
-runcasino.Estimationsofcostshavelikewiseimprovedasexperiencehasincreased.Thus,ifthebiometriciancanbegivenasatisfactoryanswertohisquery,"Whatisthequestion?"hecanbeexceedinglyhelpful.Allthestatisticalmethodscannot,ofcourse,putastudybackto-getheragainifitfallsfromthewallthroughunsteadinessatitsbiologicalbase.Attemptstocompensateforbiologicaluncertaintiesconstitutethemainreasonwhyfieldtrialsinatherosclerosisaresolargeandsocostly.Thetestsforthepresenceorabsenceofatherosclerosisarestillcrudeandtheend-pointsareinsensitive.Nothingyethastheunequivocalforceofmortalityindeterminingwhatshallbecountedintrialsinvolvingcoronaryarterydisease."Stonecoldhathnofellow,"observedWilliamFarr,thefatherofBritishepidemiology;*andimprovementinelectro-cardiographiccorrelation,angiography,orotherclinicallegerdemainhasnotyetcontradictedhim.Whiledifficultiesstillabound,thescienceofclinicaltrialsismovingslowlytowardgreaterexactness.RecentlytheHeartSpecialProjectCommitteeoftheNationalHeartInstitutehasexaminedhowcloselythegrowingtechniquesforprediction,collectionofdata,andtheman-agementoftrialswerebeingmatchedbyapplicationsfornewstudiesorthesupplementalrequeststosustainoldones.Thecommitteecon-cludedthatnewgroundrulesareneeded.Ithaslaiddownsensibleandtighterlimitsoftoleranceandsetnewstandardsforcontinuingco-ordinationandreviewforitsapprovalofsuchexperiments.Itsrecom-mendationshavethefullconcurrenceoftheNationalAdvisoryHeartCouncilanditsstaff.Toavoidtherepetitionofoldermistakestheinsti-tuteistailoringtothetimesitsmechanismsforconsiderationandman-agementofthefieldtrialsitwillsupportinthefuture.Membersofthestaff,thestudysections,andtheAdvisoryCouncilnowparticipateatearlierstagesintheprocessofapproval.ProtocolsarebeingexaminedinthelightofexperiencewithstudydesignthatIhavementioned.Specialgroupshavebeenconvenedtoconsiderthealternativestoagivendesign.Doubtandagreementarebeingcollectedanddisplayed*PerProfessorDonaldReed.Bull.N.Y.Acad.Med.99D.S.FREDRICKSON THEFIELDTRIAL99'~~~~~~~~~~~~~~~~~withheightenedawarenessoftheneedtoselectonlythosestudiesthatcanmeasureup
fullytootherresearchthatcompetesforresourcesthatpromisetofallfarshortofdemand.SomemethodsderivedfromthemanagementoftheCoronaryDrugProject,intheuseofsteeringandpolicycommittees,inthevalueofincreasedfacilityfordatacoordination,inthecentralizationoflabora-torymeasurements,willbeextendedtoothertrials.AndweshalltrytoaugmentthestaffinBethesdaasmuchaswecanforneededsupportandmaintenanceofadequatecontinuity.POPULATIONImusthastentotakeuptwofurtheraspectsofthepopulationin-gredientofthefieldtrial.Theselectionofthepopulation-at-studyhasalreadybeenmentioned.Itmaybeafewhundred;itwillprobablybemanythousandsinsometrialstocome.Whethertheyshallbewholetowns,partsofcities,free-living,"captive,"individualpatientsofin-dividualpractitioners,orparticipantsinlargercorporatemedicalsys-tems,willdependonseveralfactors.Clearly,asstudypopulationsin-creaseinsize,themorecoordinatedwilltheactionhavetobeoversuchgreatareas.BoththesystemofprovidingmedicalcareandthemobilityofAmericansmakethiscountrymoredifficultthansomeothersforconductingstudiesofgreatsize.YetanswersapplicabletoAmericanscannotalwaysbefoundabroad.Itseemsinevitablethatlargenumbersofourcitizenswillbeinvolvedinfieldtrialsofonesortoranother.Amongthoseinvariablyinvolvedarethephysicians,thesecondgroupessentialtoanyfieldtrial.Weareconcernedwithtwogroups.First,thephysicianswhoconductsuchtrials,fortheirwillingnessandtheiravailability,andnotthatofmoneyorpatients,willprobablybethefactorthatlimitsthescopeoftrialsthatcanbeattempted.Ourdebtisgreattothesemenandwomenwhoserveincollectiveresearchprojectswithrecognitionthatisnotoftencommensuratewiththeircontribution.Ihopetheywillalwaysbesustainedbyawarenessoftheiressentialityandthesenseofthepotentialforcommongoodthatmotivatesalltrialsinthefield.ButIamalsoconcernedherewiththephysicianwhoisnotengagedinthetrial,whosepatientsarepotentialsubjectsforparticipation.Whenhispatientsseekhisguidancehemusttakeaposition.Itisobviouslynotaneasyresponsibility,anditcanbeverydifficultwhenthetrialinvolvesVol.44,No.8,August1968THEFIELDTRIAL99I 992D.S.FREDRICKSONalternativesoftrea
tment.InthemiddleofthetrialoftreatmentAversustreatmentB,howoftenarewenottemptedtosingleoutpatientsforspecialconsideration,tointerferewithrandomselection?Howdiffi-cultitistoacceptthefactthatbysuchactionweareassumingthatwealreadyknowthatAorBissuperiorwhen,infact,thisdeterminationisthepurposeofthetrial.Compassionandobjectivityhaveeverbeen,andwillalwaysbe,thehornsofthephysician'sdilemma.Hecannotescapebyignoringoneortheother.Hisonlycourseistoarmhimselfwithfullknowledgeoftheelementsofcommunitydoubtandagreementthathavebroughtaboutthiscollectiveeffort.Thenhemustcometohisowndecisionandguidehispatientswithaslittlebiasaspossible.Thetotalburdencannotbebornebyafewportionsofthepopula-tion.Wealsohaveanobligationtoeducateallofsocietytoakeenersophisticationinthematterofdetermininghowonemustdeterminetheusefulnessofmeasurespertinenttohealthcare.Thedecision-makinggroupsofsocietyneedfullunderstandingoftheneedforscientifictech-niquesinlighteningtheburdensimposedbymystiqueorindifference.Theyneedtounderstandfullyourquestionandourmethods.Itisourresponsibilitytomaketheseknownclearlyandhonestly.Fieldtrialsareindispensable.Theywillcontinuetobeanordeal.Theylackglamor,theystrainourresourcesandpatience,andtheypro-tractthemomentoftruthtoexcruciatinglimits.Still,theyareamongthemostchallengingtestsofourskills.Ihavenodoubtthatwhentheproblemiswellchosenthestudyisappropriatelydesigned,andthatwhenallthepopulationsconcernedaremadeawareoftherouteandthegoal,therewardcanbecommensuratewiththeeffort.If,inmajormedicaldilemmas,thealternativeistopaythecostofperpetualun-certainty,havewereallyanychoice?APPENDIX*Inthesimplestcaseweshallbecomparinginaclinicaltrialtheyearlyincidenceofsomediseaseeventinacontrolgroup(eitherun-treatedortreatedwithanalreadyestablishedtherapy)withanothergroupsimilarinallessentialrespectsexceptthatsomenewtreatmentisgiventhissecondgroup.Anappropriatecomparisonofthetwogroups*IamindebtedtoDr.MaxHalperinandMr.JeromeCornfieldforthegistofthisappendix.Scholarswillwishtoreferatleasttothefollowingreferenceformoreillumination:Dixon,W.J.andMassey,F.J.In
troductiontoStatisticalAnalysis,2d.ed.,NewYork,McGraw-Hill,1957,pp.88-93.Bull.N.Y.Acad.Med.D.S.FREDRICKSON992 THEFIELDTRIAL993~~~~~~~~~~~~~~~thenconsistsofcomparingtheobservedratesforthespecifiedevent.Iftherateinthecontrolgroupissufficientlylargerthantherateinthetreatedgroup,weinferthatthenewtreatmentissuperiortonotreat-mentortothestandardtreatment,asthecasemaybe.Thekeytermhereis"sufficientlylarger."Thesufficiencyofproofismainlydefinedbyourchoiceoftwotermswhichthestatisticianscallaand1P.Alphaistheprobabilitythatthe"treatmenttherapy"isinferredtobesuperiortothecontroltherapywhen,infact,itisnot.Thesmallerawechoose(themorecautiouswewishtobeininferringthatthetreat-mentissuperiortothecontroltherapy),thelargertheobserveddiffer-enceinratesmustbebeforeweconcludethatthetreatmentissuperior.Onceahasbeenspecifiedanditisassumedthattherateofeventsinthetreatedgroupwill,infact,besmallerthanthatinthecontrolsbysomeminimalamount,itisusefultosetalimitfortheprobabilitythat,intheactualtrial,thetreatmentratewillnotbesignificantlysmallerthanthecontrolrate.Thislatterprobabilityiswhatthestatisticianscall1A.Thequantity(i-,8)canbethoughtofasthesensitivityofthetrial.Choiceofaand3,otherthingsbeingequal,determinesthesamplesizesofpopulationrequiredforatrial.Thesmallertheaand/8selected,thelargermustbethesizesofthetrialgroups.Letusillustratethisbyahypotheticalstudyinwhichaneventoccursinthecontrolgroupatarateintheneighborhoodofiooperthousandperyear.Furtherweexpectthenewtreatmenttodecreasethisratebyatleast30percentinoneyear.Theaccompanyingtablegivesthetotalsamplesizerequiredforaone-yeartrialforsomeselectedvaluesofaand/3.TABLE(a,p)0.01,0.050.01,0.100.05,0.050.05,0.10Totalsamplesize,withnodropout*5,4404,5003,7402,960Totalsamplesizewith1%dropout5,5004,5503,7802,990Totalsamplesizewith10%dropout6,3205,2154,3353,430Totalsamplesizewith30%dropout7,9106,5305,4304,300*Theterm"dropout"needsdefinition;asusedabove,adropoutisapersonwhogoesofftherapybutwhocanbefollowedintermsofwhetherhedidordidnotsufferaneventduringthecourseofstudy.Vol.44,No.8,August1968THEFIELDTRIAL993