臨床試驗設計的類型

Lea

Drye,

PhDJohnsHopkinsUniversityTypesofTrialDesignCopyright?2023JohnsHopkinsUniversityandLeaDrye.

AllRightsReserved.第1頁

Phase

I:First

stage

in

testing

a

new

intervention

in

humansUsually

10-30

peopleIdentify

tolerable

dose,

provide

information

on

drug

metabolism,excretion,

and

toxicityOften

not

controlled

Phase

II:Usually

30-100

peoplePreliminary

information

on

efficacy,

additional

information

onsafety

and

side

effects

Phase

III:Usually

100+

peopleAssess

efficacy

and

safetyControlled,

usually

randomized2Phases

of

trials第2頁LectureOutline

DiscussvarioustrialdesigntypesParallelCrossoverGroupallocationFactorialLargesimpleEquivalencyNon-inferiorityAdaptive3第3頁ComparisonStructure:Parallel,Crossover,andGroupAllocationDesignsSectionAThematerialinthisvideoissubjecttothecopyrightoftheownersofthematerialandisbeingprovidedforeducationalpurposesunderrulesoffairuseforregisteredstudentsinthiscourseonly.Noadditionalcopiesofthecopyrightedworkmaybemadeordistributed.第4頁ParallelDesign5

Simultaneoustreatmentandcontrolgroups

Eachpersonisrandomlyassignedtoonetreatmentgroup

Randomizationremovestreatmentselectionbiasandpromotescomparabilityoftreatmentgroups

Statisticalcomparisonsmadebetweentreatmentgroups第5頁|P叫ωe G巾Randomized「占v1，白6iJHSPH第6頁ParallelDesignExample:NETTSource:NETT

ResearchGroup(1999).Chest1999;116:

1750-61;

NETT

Research

Group

(1999).

JThoracCardiovascSurg,118:

518-528;

Fishman,

A.,&Martinez,F.,et

al.(2023).

NEnglJMed348:

2059-73.7

National

Emphysema

Treatment

Trial

(NETT)- PhaseIIItrial,unmaskedPopulationPeoplewithsevereemphysemaSamplesize1,200AllocationtotreatmentRandomizedTreatments—Lungvolumereductionsurgeryplusmedical

therapy—Medical

therapy(standardtherapycontrol)第7頁ParallelDesignExample:NETTSource:NETT

ResearchGroup(1999).Chest1999;116:

1750-61;

NETT

Research

Group

(1999).

JThoracCardiovascSurg,118:

518-528;

Fishman,

A.,&Martinez,F.,et

al.(2023).

NEnglJMed348:

2059-73.8HypothesistestingSuperiorityOutcomes—Primary:mortality,exercisecapacity—Secondary:qualityoflife,symptoms,lungfunctionandmechanics,functionalcapacityFollow-upUpto

7.5

yearsNumberofrecruitingcentersMulti-center

(17)第8頁CrossoverDesign9

RandomizationoforderinwhichtreatmentsarereceivedABorBARandomizationpromotesbalancebetweentreatmentgroupsintimingofexposure

TestingofbothtreatmentsineachpatientEachpatientservesashis/herowncontrolVariabilityreduced

becauselessvariabilitywithin

patient

thanbetween

patients

Fewerpatientsneeded第9頁CrossoverDesignGraphWashout10Group1,

Tx

AGroup1,

TxBGroup2,

TxBGroup2,

Tx

A第10頁CrossoverDesign:Disadvantages11

Treatmentcan’thavepermanent

effects

orcures

Potentialcarry-overeffectsoffirst-periodtreatmenttosecondperiodWashoutneedstobelongenoughUnequal

carry-over

effectsTreatmentduringwashout

Testforperiodbytreatmentinteractionsnotpowerful

Dropoutsmoresignificant

Analysismaybemoredifficult第11頁CrossoverDesign:Uses12

ConstantintensityofunderlyingdiseaseChronic

diseases—asthma,

hypertension,

arthritis

Short-termtreatmenteffectsReliefofsignsorsymptomsofdisease

Metabolic,bioavailability,ortolerabilitystudies第12頁CrossoverDesign:Examples13

Evening-dosevs.morning-dosedtravoprostinopen-angleglaucomafor24-hourintraocularpressurecontrol

Montelukastvs.salmeterol

as

adjuvant

to

inhaled

fluticasone

forexercise-inducedasthmainchildren

Topicaloilvs.placeboforneuropathicpain第13頁GroupAllocationDesign14

Alsoknownas

“cluster

randomization”

Randomizationunitisagroupofindividuals(community,school,clinic)

Individual

randomizationand

interventionis

notfeasibleor

isunacceptableTrackingContamination

Ifthereisacorrelationintheresponseswithinagroup,designlosessomeefficiency(moreindividualsrequired)第14頁lα[Ij川臼15Randomized〈〈第15頁Group

Allocation

Example:

Sommer

Vit

A

trial16Lancet.

1986

May

24;1(8491):1169-73

PopulationPreschool

children

in

northern