Cohort Study Design

7/25/2019 Cohort Study Design

1/25

Cohort Study DesignCohort Study Design

BLBL

August 15, 2015August 15, 2015


2/25

Learning objectiveLearning objective

At the end of the session, the students wi beAt the end of the session, the students wi beabe!abe!" #o describe and draw diagra$$atic re%resentation of#o describe and draw diagra$$atic re%resentation of

a cohort study design&a cohort study design&

" #o distinguish between retros%ective and %ros%ective#o distinguish between retros%ective and %ros%ectivecohorts&cohorts&

" #o identify factors that are considered in the seection#o identify factors that are considered in the seectionof study cohorts&of study cohorts&

" #o of data for cohort studyo of data for cohort study&" #o %erfor$ si$%e anaysis of data fro$ cohort#o %erfor$ si$%e anaysis of data fro$ cohort

studies&studies&" #o identify issues in the inter%retation of resuts of#o identify issues in the inter%retation of resuts of

anaysis of data fro$ cohort studies&anaysis of data fro$ cohort studies&


3/25

'ntroduction'ntroduction

anaytic study design that is a $oreanaytic study design that is a $ore

straightforward $ethod for drawing causastraightforward $ethod for drawing causa

inferencesinferences

#his invoves a design in which#his invoves a design in which

infor$ation about the study factor () *infor$ation about the study factor () *

e+%osure and )* non e+%osure is -nowne+%osure and )* non e+%osure is -nown

for a subjects at the beginning of thefor a subjects at the beginning of the

foow.u% %eriodfoow.u% %eriod


4/25

'ntroduction'ntroduction

#hese subjects are free of the outco$e of#hese subjects are free of the outco$e of

interest (usuay disease.free individuas ifinterest (usuay disease.free individuas if

the outco$e is incidence of disease arethe outco$e is incidence of disease are

foowed for a given %eriod throughfoowed for a given %eriod through

e+a$inations or %o%uation surveiance,e+a$inations or %o%uation surveiance,

during or after which new cases or deathsduring or after which new cases or deaths

(D are identified&(D are identified&


5/25

DescriptionDescription

#he cohort study has been caed by a#he cohort study has been caed by a

variety of na$es!variety of na$es!

" /oow.u% study/oow.u% study

" ros%ective studyros%ective study

" 'ncidence or $ortaity study'ncidence or $ortaity study

" Longitudina studyLongitudina study


6/25

Cohort sche$aCohort sche$a

C

C

) C

) C

) D

) D

) D

) D


7/25

AdvantagesAdvantages

Since one starts with individuas free of theSince one starts with individuas free of theoutco$e (disease free, the te$%ora seuenceoutco$e (disease free, the te$%ora seuencebetween the e+%osure and outco$e is $orebetween the e+%osure and outco$e is $ore

ceary estabished (one is sure that the e+%osureceary estabished (one is sure that the e+%osureoccurred before the outco$e, which is aoccurred before the outco$e, which is areuire$ent for the e+%osure to be a causareuire$ent for the e+%osure to be a causafactor&factor&

#he cohort study design aows for the direct#he cohort study design aows for the directcacuation of e+%osure s%ecific incidence rates&cacuation of e+%osure s%ecific incidence rates&

3ith cohort, one can oo- into the $uti%e effects3ith cohort, one can oo- into the $uti%e effectsof a singe e+%osure&of a singe e+%osure&


8/25

S$o-ing

Lung Cancer

Chronic bronchitis

Asth$a

4ther diseases


9/25

ote! 'n Case.Contro Studiesote! 'n Case.Contro Studies

Breast

Cancer

enarche

6istory of BC

B7CA

4ther

e+%osures


10/25

AdvantagesAdvantages

Suitabe for studies of rare e+%osures (thoseSuitabe for studies of rare e+%osures (those

identified as beonging to these grou% ofidentified as beonging to these grou% of

individuas who are e+%osed to this rareindividuas who are e+%osed to this rare

e+%osure are ta-en as a cohort grou% ande+%osure are ta-en as a cohort grou% andco$%ared to the genera %o%uation&co$%ared to the genera %o%uation&

Certain for$s of biases that occur in otherCertain for$s of biases that occur in other

anaytic designs can be $ini$i8ed, es%eciayanaytic designs can be $ini$i8ed, es%eciay

with regard to the ascertain$ent of e+%osurewith regard to the ascertain$ent of e+%osure


11/25

DisadvantagesDisadvantages

9eneray $ore e+%ensive . this is brought9eneray $ore e+%ensive . this is broughtabout by the need to ascertain theabout by the need to ascertain thedeveo%$ent of the outco$e at differentdeveo%$ent of the outco$e at different

ti$e %oints during the foow.u% %eriod&ti$e %oints during the foow.u% %eriod&Cohort studies reuire big sa$%e si8es inCohort studies reuire big sa$%e si8es instudies of rare disease&studies of rare disease&

e$bers of the cohort $ay be ost duringe$bers of the cohort $ay be ost duringthe foow.u% %eriod without -nowing theirthe foow.u% %eriod without -nowing theiractua disease status&actua disease status&


12/25

Types of Case-Control StudiesTypes of Case-Control Studies

Prospective StudiesProspective Studies

Retrospective Cohort (HistoricalRetrospective Cohort (Historical

Cohort)Cohort)


13/25

Retrospective StudiesRetrospective Studies

'n this study, a grou% or grou%s of'n this study, a grou% or grou%s ofindividuas who are free of the outco$e ofindividuas who are free of the outco$e ofinterest at so$e %oint in the %ast areinterest at so$e %oint in the %ast are

identified in ter$s of their e+%osure eveidentified in ter$s of their e+%osure eve#he cohorts $ay aso be fi+ed or dyna$ic#he cohorts $ay aso be fi+ed or dyna$ic

he outco$e status of these individuas u%he outco$e status of these individuas u%

to a s%ecified ti$e in the %ast or u% to theto a s%ecified ti$e in the %ast or u% to the%resent ti$e is deter$ined through%resent ti$e is deter$ined throughe+isting records, e&g& $orbidity or $ortaitye+isting records, e&g& $orbidity or $ortaityrecordsrecords


14/25

Retrospective StudiesRetrospective Studies

'n so$e retros%ective cohort studies, the'n so$e retros%ective cohort studies, the

foow.u% %eriod $ay s%i over to thefoow.u% %eriod $ay s%i over to the

future& So$e authors ca thisfuture& So$e authors ca this

a$bis%ective cohort studiesa$bis%ective cohort studies


15/25

Selection of the Study CohortsSelection of the Study Cohorts

de%ends on scientific considerations, feasibiityde%ends on scientific considerations, feasibiityconsiderations and freuency of e+%osureconsiderations and freuency of e+%osure" Asbestos wor-erAsbestos wor-er

" #hose who have undergone a %articuar $edica#hose who have undergone a %articuar $edicathera%y such as +.ray treat$ent or re%eatedthera%y such as +.ray treat$ent or re%eatedfuorosco%ic e+a$inationfuorosco%ic e+a$ination

" 'ndividuas iving near a sus%ected environ$enta'ndividuas iving near a sus%ected environ$entaha8ard (i&e& nucear du$%site in Car- Air Baseha8ard (i&e& nucear du$%site in Car- Air Base

" 9rou%s with unusua dietary or ifestye %ractices such9rou%s with unusua dietary or ifestye %ractices suchas the Seventh Day Adventistas the Seventh Day Adventist

" #hose who are e+%osed by being at a given event#hose who are e+%osed by being at a given eventsuch as the dro%%ing of the ato$ic bo$b atsuch as the dro%%ing of the ato$ic bo$b at

6iroshi$a&6iroshi$a&


16/25

Seection of Co$%arison 9rou%Seection of Co$%arison 9rou%

#hese co$%arison grou%s shoud be#hese co$%arison grou%s shoud be

si$iar in a res%ects to the e+%osedsi$iar in a res%ects to the e+%osed

%o%uation e+ce%t for the e+%osure%o%uation e+ce%t for the e+%osure

Another a%%roach is sa$%ing fro$ theAnother a%%roach is sa$%ing fro$ the

genera %o%uationgenera %o%uation


17/25


18/25

Anaysis of Cohort StudiesAnaysis of Cohort Studies

Data Layout


19/25

Cu$uative 'ncidence of the )+%osed (C'e * a:n1

Cu$uative 'ncidence of the ;ne+%osed (C'u * c:n2

Cu$uative 'ncidence 7atio (C'7 or 7is- 7atio (77 is!

C'7 * 77 * C'e : C'u *a:n1

c:n2


20/25


21/25

)+a$%e)+a$%e


22/25

C'e * 1>? : 1@0 * 0&


23/25


24/25

77 * 1@&>

7eject 607eject 60

5&>5 >


25/25

Issues in InterpretationIssues in Interpretation

Seection BiasSeection Bias

/oow.u% Bias/oow.u% Bias

iscassification Biasiscassification BiasConfoundingConfounding

Cohort Study Design

Documents