FDA药物流行病学-真实世界研究-课件2.ppt

1、October 2009Hangzhou1 Real-life Studies of Pharmaceutical Agents陳建煒,MD,ScD,FISPEHarvard School of Public Healthi3 Drug SafetyOctober 17-18,2009 October 2009Hangzhou2October 2009Hangzhou3Some terminology about safetylTherapeutic and Non-therapeutic use of medical productslMedication errors(adverse dr

2、ug events)and biologic effects at therapeutic dose(adverse drug reactions)lSafety signalOctober 2009Hangzhou4Safety of medical productslPatient safetylAs a result of human error/system failurelDrug/Vaccine/Device safetylToxicity occurred when the product was used appropriatelylGray arealInappropriat

3、e use(not only off-label use)October 2009Hangzhou5By Dr.Robert Nelson,in Hartzema,Tilson,ChanOctober 2009Hangzhou6Methods and objectivesInterventionalObservational studies(clinicalstudies trials)BenefitEfficacy(Comparative)Phase II/IIIEffectivenessRiskPragmatic trialReal-life use ofLarge&Simple tria

4、lmedical productsOctober 2009Hangzhou7Applications of pharmacoepidemiologylSafety assessmentlRelative and absolute risk estimateslRisk factors for adverse outcomeslUsage patternslSafety signal detectionlComparative effectivenesslDrug utilizationl(Development of medical products)October 2009Hangzhou8

5、Three examples for today and tomorrowlStatins and rhabdomyolysislRisk factors for Stevens Johnson SyndromelAprotinin in cardiac surgery and increased risk of end-organ damage October 2009Hangzhou9lWhy?lPharmacoepidemiologyl(Therapeutic)risk managementOctober 2009Hangzhou10Journal of the National Can

6、cer Institute 2005;97:872-3l“The real problem in safety is often not in the drug but how it is used,”-Raymond Woosley,MD,PhD,Critical Path Institutel“In many ways,after a drug is released,the real trial begins,”-Paul Seligman,MD,MPH,Office of Pharmacoepidemiology and Statistical Science,FDAOctober 2

7、009Hangzhou11A Brief review:from laboratory to bedside(T1 and T2)lFrom laboratory to approval(FDA)lTranslational researchlhttp:/www.fda.gov/oc/initiatives/criticalpath/lhttp:/www.c-path.org/lFrom approval to bedside/community(Agency for Healthcare Research and Quality AHRQ)lSometimes called diffusio

8、n of technologylThe role of epidemiology in this processOctober 2009Hangzhou12R&D in the industrylResearchlBasic/bench researchlAnimal,cellular,and molecular modelslDevelopmentlBring a molecule to the marketlEvaluation from clinical and business perspectiveOctober 2009Hangzhou13Clinical developmentl

9、The role of surrogate markers and surrogate end pointslSolid epidemiology data are required to substantiate the correlation between surrogate end points and important clinical outcomeslEnd-point trials and Outcome trialslMay or may not be good enough for safety assessmentOctober 2009Hangzhou14Clinic

10、al developmentlPhase IIIb and IVlNew dosage and formulation(supplementary NDA)lNew indications(supplementary NDA)lLong term outcomeslEffectiveness(vs.efficacy)lSafetylQuality-of-life/Functional status/Cost-effectiveness outcomesOctober 2009Hangzhou15Whats wrong with the current system from a safety

11、perspective?lLimited generalizability of animal,cellular,and molecular modelslhttp:/content.nejm.org/cgi/content/abstract/333/17/1099lLimited utility of surrogate end pointslFor effectivenesslFor safetylInsufficient knowledge about background rates and risk factors for adverse outcomesOctober 2009Ha

12、ngzhou16Whats wrong with the current system from a safety perspective?lLimitations of Phase III trialslComparison grouplDurationlPatient populationlPre-marketing vs.Post-marketinglClinical trials vs.Observational studieslThe challenge of combining studies(meta-analysis)lMisuse of statisticsOctober 2

13、009Hangzhou17The Science of Safety“We are seeing the emergence of a science of safety.This science combines the growing understanding of disease and its origins at the molecular level(including understanding of adverse events resulting from treatment)with new methods of signal detection,data mining,

14、and analysis,enabling researchers to generate hypotheses about,and confirm the existence,and causal factors,of safety problems in the populations using the products.”October 2009Hangzhou18http:/www.fda.gov/Safety/FDAsSentinelInitiative/ucm089474.htmOctober 2009Hangzhou19Science of Efficacy and Scien

15、ce of SafetylDifferent paradigmlTwo sets of related methodology,data sources,interpretation of findings,and regulatory implicationslScience of efficacylHypothesis testinglControlled experimentslSingle end point/outcomelScience of safetylRisk identification/Signal detection(hypothesis generation)lPub

16、lic health surveillancelRisk assessment(hypothesis testing)lExperimental(clinical trial)and observational(epidemiology)methodsOctober 2009Hangzhou20Science of SafetylRisk identification/signal detectionlHypothesis generationlRisk assessment/signal confirmation or refutationlHypothesis testinglRisk m

17、itigation/managementOctober 2009Hangzhou21Data sources for safety signalslAnimal/cellular modellClinical trials(pre-and post-marketing)lAdverse eventslLaboratory data(as surrogate markers)lNumerical imbalance vs.statistical significancelSpontaneous reports(Pharmacovigilance)October 2009Hangzhou22Spo

18、ntaneous Adverse Drug Reactions Reporting SystemslVoluntarylFour elementslAn identifiable patientlAn adverse eventlA suspected drug/vaccine/devicelAn identifiable reporterlSensitivity over SpecificityOctober 2009Hangzhou23Serious Adverse Eventslhttp:/www.fda.gov/medwatch/report/DESK/advevnt.htmlDeat

19、hlLife threateninglHospitalization(initial or prolonged)lDisabilitylCongenital anomalylRequires Intervention to Prevent Permanent Impairment or DamageOctober 2009Hangzhou24Strengths of the spontaneous reports systemlModerate costslA case or case series can tell a compelling story(temporal sequence,d

20、e-challenge&re-challenge)lMinimal delaylWorks well with public health-oriented clinicians October 2009Hangzhou25LimitationslAnecdotal informationl“The plural of anecdote is not data”l(Usually)poor data qualitylCannot conclusively establish causalitylReporting rates affected by many secular factorslH

21、ypothesis generation,not confirmatoryOctober 2009Hangzhou26Hypothetical case#1Sudden deathlYear 1988,62 year-old male,no CV risk factor,survived acute MIlFrequent premature ventricular complexes(PVCs)after MIlPatient was given encainide,a class Ic anti-arrhythmic agent,to suppress the PVCsl6 months

22、later,patient died suddenlylBackground rate:one-year mortality 10%October 2009Hangzhou27Would anyone suspect that death was precipitated by the drug?lWhat if there was no CAST trial?lLeary WE.Warning Issued on 2 Heart Drugs After Deaths of Patients in Test.New York Times,April 26,1989,page 1A.lCardi

23、ac Arrhythmia Suppression Trial(CAST)Investigators.Preliminary report:N Engl J Med 1989;321:406-12.(Aug 10 issue)October 2009Hangzhou28Hypothetical case#2Liver failurelYear 1999,53 year-old male,BMI 31,social drinker,on glyburide,troglitazone,atenolol,simvastatin,and diclofenaclPatient developed acu

24、te liver failure and died while waiting for liver transplantationlLiver failure may be caused by a druglbut which drug or combination of drugs?October 2009Hangzhou29What is the difference between case#1 and case#2?lSuspicion and reporting of Adverse Drug ReactionslSeverity of eventlPrior knowledgelA

25、lternative explanationslBackground rate of the adverse eventlRarelNot so rareOctober 2009Hangzhou30Rare and not-so-rare adverse eventslRarelRhabdomyolysis,DILI,Stevens-Johnson Syndrome,agranulocytosis,Churg-Strauss Syndrome,lNot so rare(especially among certain populations)lMyocardial infarctionlTub

26、erculosis among RA patientslChronic active hepatitis B exacerbationlSuicide among adolescentsl,October 2009Hangzhou31Hypothetical case#3Acute myocardial infarctionlYear 2000,49 year-old male,smoker,BMI 33,mild hypertension and LDL elevationlLeft knee osteoarthritislGiven rofecoxib 25mg/day for two w

27、eeks,then the dose increased to 50mg/dayl9 months later,had non-fatal myocardial infarctionlBackground rate of AMI:several cases per 1,000 person-years October 2009Hangzhou32An interesting questionlCould the existing system identify CV safety signals associated with rofecoxib if there was no VIGOR t

28、rial?lPUBMED search on Oct 24,2005,published case reports of rofecoxib:hepatitis,interstitial nephritis,colitis,angioedema,anaphylactic shock,gynecomastia,acute renal failure,delirium,Stevens-Johnson syndrome,congestive heart failure,transient visual impairment,aquagenic wrinkling of the palms,Octob

29、er 2009Hangzhou33Epidemiology,a general outlinelMeasures of Disease OccurrencelIncidence and PrevalencelCumulative incidence and incidence ratelConcept of dynamic population and person timelMeasures of AssociationlRelative risklRate ratiolOdds ratioOctober 2009Hangzhou34Epidemiology,a general outlin

30、elDescriptive EpidemiologylCase reportslCase serieslFrequency of eventslStudies providing some evidence,but not definitivelCross-sectional studieslEcological studiesOctober 2009Hangzhou35Epidemiology,a general outlinelObservational studies,selected methodslCohort studieslCase-control studieslCase-cr

31、ossover studieslInterventional studieslClinical trials for individual subjectslClustered randomized trialsOctober 2009Hangzhou36Epidemiology,a general outlinelCombining information from multiple studieslMeta-analysislData poolingOctober 2009Hangzhou37Systematic and random errorslValidity considerati

32、on in epidemiologylConfoundinglBiaslInformation biaslSelection biaslAssessment of random error in epidemiology:statisticsOctober 2009Hangzhou38Commonly used terms in epidemiologyConfounder(s)ExposureOutcomeOctober 2009Hangzhou39Exposures in pharmacoepidemiologylDrugslVaccineslMedical deviceslProcedu

33、reslProgramsle.g.screening programslCoding systems for each of themOctober 2009Hangzhou40Ascertainment of OutcomeslOperational definitionslSensitivity and specificitylObjective criterialE.g.hemorrhagic strokelBlinded adjudication if possibleOctober 2009Hangzhou41Performance of ICD-9 codes in the stu

34、dy of rhabdomyolysisOctober 2009Hangzhou42Andrade et al.Health plan administrative databases can efficiently identify serious myopathy and rhabdomyolysis.J Clin Epi 2005;58:171-4October 2009Hangzhou43Cohort studies in pharmacoepidemiologylSource populationlAscertainment of exposurelComparison groups

35、/cohortslConfounding?lAscertainment of outcomeslInformation Bias?lControl for confoundinglEvaluation of effect modificationlStrengths and limitationsOctober 2009Hangzhou44Exposure and OutcomelExposures of interest/Comparison groupslHead-to-head comparisonlDrug vs.Disease comparisonlMultiple comparis

36、on groupslOutcomes of interestlClinically relevantOctober 2009Hangzhou45Arch Intern Med 2002;162:936-42.lThree study groupslAlendronate users(n=6,432)lWomen with osteoporotic fracture(n=9,776)lAge-gender matched unexposed(n=33,176)lComparing alendronate users and unexposedlAdjusted rate ratio=1.8,95

37、%confidence interval 0.8 3.9lComparing alendronate users and women with fracturelAdjusted rate ratio=1.1,95%confidence interval,0.6-2.2October 2009Hangzhou46Defining cohortslMatchinglIndividually-matchlFrequency-matchlMatching in cohort study vs.matching in case-control studylRestriction/Selection c

38、riterialNew-user designlIncident users vs.Prevalent usersOctober 2009Hangzhou47Bias and confoundinglConfoundinglChanneling,Confounding-by-Indication,and Confounding-by-SeveritylHow to measure indication and severity?lOther known risk factors for the outcome of interest(age,gender,smoking)lBiaslDetec

39、tion bias as a form of information biasOctober 2009Hangzhou48Analytic approacheslAt risk person-timelImmortal person-timelCumulative incidence,incidence,and correlated outcomeslCount,person-time,and time-to-eventlControl for confoundinglMultiple regression,propensity score,instrumental variables,and

40、 otherslAccount for unmeasured confounderslNested case-control studyOctober 2009Hangzhou49Stratified analysislControl for confoundinglEffect modificationlSpecific subgroup(s)lDrug-drug interactionOctober 2009Hangzhou50Cohort studies in pharmacoepidemiologylStrengthslRelative risk and Absolute risk e

41、stimateslMultiple outcomes(safety surveillance for multiple outcomes and risk-benefit assessment)lLimitationslMitigated by the availability of secondary and hybrid data sourcesOctober 2009Hangzhou51Clinical trialslClinical Regimen trials vs.Clinical practice trialslInclusion and Exclusion criterialT

42、itration,switching,and stoppinglFollow-up/monitoringlNew Drug Application vs.Clinical practicelPlacebo-controlled vs.head-to-head comparisonlInternal validity vs.GeneralizabilitylComorbidity and potential drug-drug interactionOctober 2009Hangzhou52Explanatory vs.PragmaticRoland and Torgerson.Underst

43、anding controlled trials:What are pragmatic trials?BMJ 1998;316:285“Trials of healthcare interventions are often described as either explanatory or pragmatic.Explanatory trials generally measure efficacythe benefit a treatment produces under ideal conditions,often using carefully defined subjects in

44、 a research clinic.Pragmatic trials measure effectivenessthe benefit the treatment produces in routine clinical practice.”October 2009Hangzhou53Some large trialslALLHAT studylhttp:/allhat.uth.tmc.edu/lPROVE-IT/TIMI-22 studylhttp:/content.nejm.org/cgi/content/abstract/350/15/1495lCATIElhttp:/www.nimh

45、.nih.gov/press/catie_release.cfmlhttp:/content.nejm.org/cgi/content/abstract/353/12/1209October 2009Hangzhou54Clinical trials for safetylLarge and Simple TriallLarge and Dirty trial(Meinert)lA hybrid approachlRandomizationlIndividual levellGroup levellReal-life management and follow-uplImplications

46、for comparative effectiveness researchOctober 2009Hangzhou55Example of a large and simple triallhttp:/clinicaltrials.gov/ct2/show/NCT00418171lLarge Simple Trial(LST)Of Cardiovascular Safety Of Ziprasidone And Olanzapine-(ZODIAC)lAn International,Multicenter,Large Simple Trial(LST)To Compare The Card

47、iovascular Safety Of Ziprasidone And OlanzapinelThe primary objective of the study is to estimate the relative incidence among users of ziprasidone and olanzapine of non-suicide mortality within 12 months.l18,239 enrolled patientsOctober 2009Hangzhou56Eligibility criteria for ZODIAClInclusion Criter

48、ia:lPatients newly treated for schizophrenia and those receiving continuing treatment will be eligible if the treating psychiatrist is ready to initiate a new antipsychotic medication,and would consider using either ziprasidone or olanzapine as an appropriate therapy.lAge 18 and olderlMale and femal

49、e patients who meet all criteria listed below are eligible to be enrolled in this study:lDiagnosed with schizophrenia lWilling to provide information on at least one alternate contact person for study staff to contact regarding patients whereabouts,should the patient be lost-to-follow-up over the co

50、urse of the study lExclusion Criteria:lProgressive fatal disease of a life expectancy which prohibits them from participating in a one year research study lPreviously randomized to study medication and enrolled in this study October 2009Hangzhou57Clinical trial,statistical perspectivelOne(and only o