1.Foundation of clinicaltrialsAssociationdoesNOTimplycausation:Observationalstudieshavesuggestedthatlowserumbeta-caroteneisassociatedwithanincreaseinlungcancerContrarytotheobservationalstudies,twoclinicaltrials,(theAlpha-Tocopherol,Beta-CaroteneCancerPreventionStudyGroup,1994)and (Omenn etal.1994),demonstratedthatthe incidenceof lung cancerincreased inthosegivenbeta-carotenesupplementsdespitethedocumentedincreaseinserumbeta-carotene.Inathirdtrial (Hennekensetal.1996),serumbeta-carotenewasincreasedbuttheincidenceof lungcancerandall cancersdidnotchange.Remarkably,theobservationofanassociationbetweenthebaseline level of beta-carotene and the incidence of lungcancerwasfoundinallthreetrials.ReplicationdidnotNIVERSITY福guaranteethattheassociationis causal.IMANITORKikui Wang (Universityof Manitoba)Statistical Decision Analysis201319/198
1. Foundation of clinical trials Association does NOT imply causation: Observational studies have suggested that low serum beta-carotene is associated with an increase in lung cancer. Contrary to the observational studies, two clinical trials, (the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study Group, 1994) and (Omenn et al. 1994), demonstrated that the incidence of lung cancer increased in those given beta-carotene supplements despite the documented increase in serum beta-carotene. In a third trial (Hennekens et al. 1996), serum beta-carotene was increased but the incidence of lung cancer and all cancers did not change. Remarkably, the observation of an association between the baseline level of beta-carotene and the incidence of lung cancer was found in all three trials. Replication did not guarantee that the association is causal. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 19 / 198
1.Foundation of clinical trialsAssociation does NOT implycausation!Causationcanbeestablishedonlyindesigned experimentsTheexperimentalapproachtoclinical research:Aclinicaltrialisthemostdefinitiveandeffectivemethodtodeterminewhetheranewmedical interventionhasthepostulatedeffects,andisalsoaneffectivemechanismtodistinguishincidenceofsideeffectsandadverseeffectsduetotheinterventionfromthosecausedbythediseaseprocessitself.Somewidelyusedbutuntestedinterventionswereultimatelyshowntobeineffectiveorevenharmfulbyclinicaltrials.Theneedforclinicaltrials:Theexampleoffreezingthestomachforduodenal ulcershowedineffectivetreatment(Ruffinetal1969).脑JNIVERSITLMANITOAAXikui Wang (University of Manitoba)StatisticalDecision Analysis201320/198
1. Foundation of clinical trials Association does NOT imply causation! Causation can be established only in designed experiments. The experimental approach to clinical research: A clinical trial is the most definitive and effective method to determine whether a new medical intervention has the postulated effects, and is also an effective mechanism to distinguish incidence of side effects and adverse effects due to the intervention from those caused by the disease process itself. Some widely used but untested interventions were ultimately shown to be ineffective or even harmful by clinical trials. The need for clinical trials: The example of freezing the stomach for duodenal ulcer showed ineffective treatment (Ruffin et al 1969). Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 20 / 198
1Foundation of clinical trialsTheneedforclinicaltrials:Hormonereplacementtherapy(HRT)isusedtotreatsymptomsinpostmenopausalwomenandisalsoknowntoreducebonelossinthesewomen,leadingtoreducedbonefracturerates.HRTalsoreducesserumcholesterolleadingto the belief that it should also reduce cardiovascularmortalityandmorbidity.Further,largeobservationalstudieshaveshownlowercardiovascularmortalityforwomenusingHRT than those not using HRT (Barrett-Connor and Grady,1998).TheseobservationsledtoawidespreaduseofHRTforthepreventionofcardiovascularmortalityandmorbidity as wellas other indications.UNIVERSITY福LMANITOSAXikui Wang (University of Manitoba)Statistical Decision Analysis201321/198
1. Foundation of clinical trials The need for clinical trials: Hormone replacement therapy (HRT) is used to treat symptoms in postmenopausal women and is also known to reduce bone loss in these women, leading to reduced bone fracture rates. HRT also reduces serum cholesterol leading to the belief that it should also reduce cardiovascular mortality and morbidity. Further, large observational studies have shown lower cardiovascular mortality for women using HRT than those not using HRT (Barrett-Connor and Grady, 1998). These observations led to a widespread use of HRTfor the prevention of cardiovascular mortality and morbidity as well as other indications. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 21 / 198
1Foundation ofclinical trialsTheneedforclinicaltrialsTwotrialsevaluatedthebenefits ofHRTinpostmenopausalwomen:onetrial inwomenwithexistingcardiovasculardisease andthe second without any evident disease.Thefirstdemonstratednobenefitandsuggestedapossibleriskof thrombosis (i.e., blood clots) (Grady etal.1998).Thesecond demonstrated a harmful effect due toblood clottingandnocardiovascularbenefit(WritingGroupfortheWomen'sHealthInitiativeRandomizedClinicaltrial,2002)ThesetrialsdiminishedtheuseofHRTforpurposesofreducingcardiovasculardisease,butHRTisstill usedforshort-termsymptomreliefinpostmenopausalwomen.福UNIVERSITYLMANITOAXikui Wang (University of Manitoba)Statistical Decision Analysis201322/198
1. Foundation of clinical trials The need for clinical trials: Two trials evaluated the benefits of HRT in postmenopausal women: one trial in women with existing cardiovascular disease and the second without any evident disease. The first demonstrated no benefit and suggested a possible risk of thrombosis (i.e., blood clots) (Grady et al. 1998). The second demonstrated a harmful effect due to blood clotting and no cardiovascular benefit (Writing Group for the Women’s Health Initiative Randomized Clinical trial, 2002). These trials diminished the use of HRT for purposes of reducing cardiovascular disease, but HRT is still used for short-term symptom relief in postmenopausal women. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 22 / 198
1Foundation ofclinical trialsTheneedforclinicaltrialsAclassofdrugsknownasbeta-blockerswasknowntobeeffectiveforloweringbloodpressureandreducingmortalityinpatientssufferingheartattacks.Scientistsbelievedthatthesedrugsshouldnotbeusedinpatientswithheartfailurebecauseinthesepatients,theheartdoesnotpumpbloodefficiently.Itwasbelievedthatloweringthebloodpressureandheartratewouldmaketheproblemsevenworse.Butaseriesoftrials(TheInternationalSteeringCommitteeonBehalf oftheMERIT-HFStudyGroup,1997;Packeretal.2001;Krumetal.2006)demonstratedconvincinglyanapproximate3o%reductionincardiovascularmortality.Aneffectivetreatmentwas ignoredformorethanadecadebecauseofbeliefinamechanistictheorywithoutclinicalUNIVERSIT囍evidence.IMANITOAXikui Wang (University of Manitoba)StatisticalDecisionAnalysis201323/198
1. Foundation of clinical trials The need for clinical trials: A class of drugs known as beta-blockers was known to be effective for lowering blood pressure and reducing mortality in patients suffering heart attacks. Scientists believed that these drugs should not be used in patients with heart failure because in these patients, the heart does not pump blood efficiently. It was believed that lowering the blood pressure and heart rate would make the problems even worse. But a series of trials (The International Steering Committee on Behalf of the MERIT-HF Study Group, 1997; Packer et al. 2001; Krum et al. 2006) demonstrated convincingly an approximate 30% reduction in cardiovascular mortality. An effective treatment was ignored for more than a decade because of belief in a mechanistic theory without clinical evidence. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 23 / 198