> 0(
fb/0DTimes New Roman4'v0(
0DTimes New Roman (Arabic)(
0
`.
@n?" dd@ @@``H$.5&'
T ' ')& nqtw z}
" $&(*0
5,.c$@3f@f,g4CdCdv0
ppp@<4BdBd`
0L
'g4+d+dv0ppp@uʚ;2Nʚ;<4!d!d{
0<4dddd{
0:2___PPT9/0?%O5'!)Potential Errors In Epidemiologic Studies**<8 zLearning Objectives"HB Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB (Prevent confounding
Evaluate confounding")(6() ConfoundingH nA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outcome and can even change the direction of the observed effect
0!0*v 2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H* Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $0,8
kRestriction Strength H`*
=Straightforward
Convenient if criteria are narrow
Inexpensive>><> Restriction Limitation,<
BB8
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor, l Randomization B*
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization, m Randomization Strength(B8 Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor,* Randomization Limitation(B8
.Not easy to perform
Ethical problems
Expensive//B/ nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data,*o Q oMatching StrengthH*
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0f Matching Limitation(B8
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0 pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B<* Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor, Stratification LimitationB*
JDifficult if many variables with varying number of categories are requiredK!K,K }Multivariate Analysis< yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength< zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <! TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!, /67BCDEFHIJKLMNOPQRTUVWXY[ ` ̙33` ` ff3333f` 333MMM` f` f` 3>?" dd@,?" dd@ " @ ` n?" dd@ @@``PR @ ` `p>> (
6
' P'
X Click to edit Master title style!!
0X' '
RClick to edit Master text styles
Second level
Third level
Fourth level
Fifth level!
S
0P]' ``'
T*
0' ` '
V*
0C' ` '
V* H
0h ? $Blank Presentation0P&(
0

V*
0 
X* d
c$ ?Dx
0X 
RClick to edit Master text styles
Second level
Third level
Fourth level
Fifth level!
S
6 
V*
6@ 
X* H
0h ? ̙338`( q
0Hz'
'
V*
0}' '
X*
6( '
V*
6 '
X* H
0h ? ̙33
C0
xV(
x
x
Cxfyd)?0<P00`02
x
S~Cfyd)?0<PP
]IV.(<< l
x CdEP

B
x
S~Jfyd)?0<P
W
Confounding <
x
<M
vDr. Sherine Shawky<* H
x0h ?
7B:p(
&
3ºyd@) ?002<P
l
C8ú
H
0h ?
6TL(
2
Sfyd@) ?002<P
r
S0`
H
0h ?
D~( PX
P
܆'fyd@) f,k,kXX?002<P
'
Cd''
&xYaH
0h ?
E<(
'fyd@) '
&xYaH
0h ? Bph
h( Ow@
h,
h C<'fyd@) ?002<P'
h
<p'PL 0
ZExposure 6,
h
<h'`
]Confounding 6,
h
<'0
OOutcome6 R
h
s*0 pR
h
Hz,ff)X?
p
h
H4 ff)X? p
h
H
33XX?0
Z p
h
H 33XX?0
[ H
h0h ? ̙33V
F(
&
3'fyd@) ?002<P
'
S~'fyd)f,k,kXX?`
]Prevent$6,
S~'fyd)f,k,kXX?`0
[Study$6,
S~'fyd)f,k,kXX?``
^Evaluate$ 6, H
0h ?
W0`( #\
D
t'fyd@) ?002<PP
'
R
3r'd?0<P
lMatching$ 0, R
3r'd?0<P
oRestriction$0, R
3r'd?0<P
B,
<,'
'
U
Randomization0 H
0h ? HNFP( <@ܙ
,
C'fyd@) ?002<P '
r
S''
H
0h ? ̙33
Kf^p( 00ح
D
`Wfyd@) ?002<PP
r
S]
H
0h ? ̙33
Lh`(
,
Cgfyd@) ?002<P
Shm
&x~(aH
0h ? ̙33INF( DHo
,
C@zfyd@) ?002<PPp`
r
Sz
H
0h ? ̙33
MNF( 48<
,
CXfyd@) ?002<P
r
S
H
0h ? ̙33
NNF( (q@
,
Cfyd@) ?002<P
r
SL0`
H
0h ? ̙33JNF(
,
Cfyd@) ?002<P
r
SD
H
0h ? ̙33
ONF0(
,
Cfyd@) ?002<P
r
S(`
H
0h ? ̙33
PNFP(
,
Cfyd@) ?002<P
r
Sи
H
0h ? ̙33.
Qpn( oqx
,
Chfyd@) ?002<P`
$3) X)?0y2P< 0
H
0h ? ̙33RNF( oqx
,
Cfyd@) ?002<P
r
SdP
H
0h ? ̙33
YNF( z w˟
,
Cfyd@) ?002<P
r
S
H
0h ? ̙33
[h`(
,
C\fyd@) ?002<P`
SP0P
&xxxhH
0h ? ̙33
T\Tp(
p&
p 3fyd@) ?002<P Pp

p CP0
&xYaH
p0h ?
Uh`(
,
Cfyd@) ?002<P

SH
&xYaH
0h ?
Vh`0( h
,
Cfyd@) ?002<P

S0

&xYaH
0h ?
X`X t(
t&
t 3xfyd@) ?002<P P
t C8
*e!)1H
t0h ? {0(
R
3PP
'
C'P'
" H
0h ? ̙33z0B:( Xv@`b@
R
3 @
t@
C4 @`
hThis lecture aims to show the epidemilogic thinking process for preventing and evaluating confounding.
ih8
> H
0h ? ̙33M0
l(
l^
l S` '
l c$Ⱥ'
0'
The confounding factor must be associated with the exposure. The confounder, independent of the exposure, is a risk factor for the outcome. The observed relationship between the exposure and outcome can be attributed totally or in part to the effect of the confounderF@
H
l0h ? ̙33P0@(
R
 3` 
 C 
Sherine Shawky, MD, Dr.PH
Assistant Professor
Department of Community Medicine and Primary Health Care
Faculty of Medicine and Allied Sciences
King Abdulaziz University, Jeddah, Saudi Arabia
shshawky@hotmail.com
z
! H
0h ? ̙330(
R
3P
'
C'p0'
" H
0h ? ̙330 (  .
X
C '
SĢ'
'
" H
0h ? ̙330 "( x@8@
R
3` '
C'
'
0Efforts are directed towards preventing the confounding effect before initiating the study and evaluating its impact on the study findings.
F8 H
0h ? ̙33Xj0`(
R
3 ` t
CUP
`Restriction is used to control confounding in the design stage for all analytical study designs.aa` H
0h ? ̙33Rl0
(
R
3 P t
C
ZRandomization is used to control confounding in the design stage for intervention studies.[[Z H
0h ? ̙33Qn0 ( >E
R
3` D
Cإ
YMatching includes elements in both design stage and analysis for all analytical studies.ZZY H
0h ? ̙33k0(
R
3 P
t
CcP@
" H
0h ? ̙330( w@8@
R
3 P
t
CvP
" H
0h ? ̙33m0( L1LL
R
3
D
C @
" H
0h ? ̙330(
R
3
D
C 0
" H
0h ? ̙33o0@(
R
3
D
CP@`
" H
0h ? ̙330`( T\
R
3 p
t
CP
" H
0h ? ̙33p0)(
R
3 D
Cx P
0
7In all analytical studies, confounding must always be considered as an alternative explanation for the study findings. Stratification and multivariate analysis are two methods used to control the effect of confounding in the analysis stage H
0h ? ̙33q0c( q@i@
R
3` D
CD
qStratification is a technique in which data are stratified by the levels of confounding factor and the relative risk estimates are compared between the different strata. If the effect is thought to be uniform in all strata, the MantelHaenszel relative risk (RRMH) is computed. Hypothesis testing is performed on the unconfounded estimate using the MantelHaenszel chisquare and the confidence interval is calculated.
,
~ A J
7 H
0h ? ̙33}02(
R
3
t
C
@H
0h ? ̙330 >(
X
C
t
Sl

@H
0h ? ̙330@>(
X
C
t
SӺ
@H
0h ? ̙330@k( t
R
3P D'
Cx'P
'
y1Three methods may be used to prevent confounding.222 H
0h ? ̙33~0P( 5@
R
3
t
Cں p
" H
0h ? ̙33v0(
R
3
C
" H
0h ? ̙330( 0
R
3 t
C

" H
0h ? ̙33r009f?e'hmKpu=x}e IK 3Sˈ
CbQN/QY^`8jrzK{G?Y_
lt/(
fL/0DTimes New Roman(b(b4bj0XbXb`0DTimes New Roman (Arabic)bXb`0`
@n?" dd@
Oh+'0t@
px ,
LXd
p!Potential Errors In EpidemiologyMiDr. Sherine Shawky CC:\Program Files\Microsoft Office\Templates\Blank Presentation.pot m F245Microsoft PowerPointoso@Y2@Oå@(b_@XiG>oM l& &&#TNPP`0D
v
&
TNPP &&TNPP
 " ! "&G&
 &Gy& &v &$ww$w&&&& $ ,$ " A$" 3 U"$3 D k*$D U 3$U f <$f w D$w L$ !R$! 2X$2 C\$C T`$T eb$e vd$v e$ &&& "&$&8L&&&
&& && && $ ,
$ " A
$" 3 U"
$3 D k*
$D U 3
$U f <
$f w D
$w L
$ !R
$! 2X
$2 C\
$C T`
$T eb
$e vd
$v e
$ &
&&&&<$4 '!#<$4 2 !#<$2 0 !<$0 . =$. , =$, ) >$) '!?$'!"
?$'!%!?$%!#!?$#!!!?$!!"@$""@$""
A$"%
A$"#
A$##A$#$B$$$B$$$C$$%D$%(D$%
%D$
%&D$& 'D$ ''E$'(E$((F$(F$()F$)*F$*+G$+,G$,,H$,I$2!I$.I$./I$/0I$01 J$12! J$22!!K$29(!K$23#!K$35$#K$56%$K$67&%L$78'&L$89('M$9@/(M$9:)(M$:;*)M$;=,*N$=>,N$>@/O$@G6/O$@A0/O$AC20O$CD32P$DF53P$FG65Q$GP?6Q$GI86Q$IK:8Q$KL;:R$LN=;R$NP?=S$PYH?S$PRA?S$RTCAT$TVECT$VYHEU$YbQHU$Y\KHU$\_NKU$_bQNV$bl[QV$bdSQV$dgVSW$giXVW$il[XX$lve[X$lq`[X$qve`X$vpeY${pY${Y$X$X$X$X$X$W$V$V$U$U$S$Q$G@{/6O$@9r({/M$9y2h!r(K$y2o^h!K$y2t0ch!J$t0o^cI$od(S^I$ok,[^H$k,h*W[G$h*d(SWF$d(Y%HSF$d(`'OSE$`']&LOD$]&Y%HLD$Y%M"<HD$Y%V$EHC$V$S$BEB$S$P#?BA$P#M"<?A$M"A!0<A$M"K":<@$K"H"7:@$H"F!57?$F!C!25?$C!A!02?$A!4 #0?$A!> 0>$> < +=$< 9 (+=$9 7 &(<$7 4 #&&&&&e$ Gb;d
$;b}Vb
$V}q`
$q\
$W
$R
$K
$ D
$ ;<
$;V/3
$/VqJk*
$JqeU"
$e?
$+
$
$&&&&$@4 '!"%(29@GPYblv'4AMYdoyvlbYPG@9y2od(Y%M"A!&8L&&&
&&
&&$@4 '!"%(29@GPYblv'4AMYdoyvlbYPG@9y2od(Y%M"A!&&e$ Gb;d
$;b}Vb
$V}q`
$q\
$W
$R
$K
$ D
$ ;<
$;V/3
$/VqJk*
$JqeU"
$e?
$+
$
$&
&& Times New Roman̞ .
2
IV:.Times New Roman̞ . 2
?..y@H Times New Roman̞ .$2
iPotential Errors In1(#(5#$(# ,. .2
f
Epidemiologic5#C((($. .2
O Studies,,$.&6)&$pH7_$p_&&&&$HH%+
$H%!HJ?
$HJ!FHoT!
$HoFkHj*
$HkH3
$HH<
$HHD
$HH)K
$H)%HNR
$HN%JHsW
$HsJoH\
$HoH`
$HHb
$HHd
$HHe
$H)HR&&&&$pHp(&8L&&&
&&
&&$pHp(&&$HH%+
$H%!HJ?
$HJ!FHoT!
$HoFkHj*
$HkH3
$HH<
$HHD
$HH)K
$H)%HNR
$HN%JHsW
$HsJoH\
$HoH`
$HHb
$HHd
$HHe
$H)HR&
&& .2

Confounding :(,(,,(.@ .
2
Dr. :#. .2
'Sherine Shawky,$#,$,(9(."System&TNPP & $BisLp $Ls}Vh) $V}``& $`jX# $jtP  $t~H $~sLearning ObjectivesPerformance ObjectivesConfounding Criteria for ConfoundingControl of ConfoundingPrevention of Confounding
RestrictionRestriction Strength Restriction Limitation Randomization Randomization Strength Randomization Limitation MatchingMatching Strength Matching LimitationEvaluation of ConfoundingStratified AnalysisStratification StrengthStratification LimitationMultivariate AnalysisMultivariate Analysis Strength!Multivariate Analysis LimitationConclusion n (Arabic)bXb`0`
@n?" dd@ @@``H$.5"!
T "!
!nqtw z}" !#%(
+&'*,./2( $501c$@3f@fg4!d!ddb`0\bx ~ppp@<4BdBd#(bg4+d+ddb`0\bppp@<4!d!d#(b:2___PPT9/0?%O5)Potential Errors In Epidemiologic Studies**<*
zLearning Objectives"HB
Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB
(Prevent confounding
Evaluate confounding")(6(
)ConfoundingH
nA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outcome and can even change the direction of the observed effect
0!0 v2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H" Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $0,,
kRestriction Strength H`
=Straightforward
Convenient if criteria are narrow
Inexpensive>><
> Restriction Limitation,<
BB,
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor,
l Randomization B"
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization,
m Randomization Strength(B, Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor, Randomization Limitation(B,
.Not easy to perform
Ethical problems
Expensive//B
/nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data, oQ oMatching StrengthH
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0
f Matching Limitation(B,
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0
pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B< Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor,
Stratification LimitationB
JDifficult if many variables with varying number of categories are requiredK!K,
K}Multivariate Analysis<
yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength<
zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <
!TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!,
/67BCDEFHIJKLMNOPQRTUVWXY[
Lh`(
,
C
fyd@) ?002<P0
@@``H$.&'
T ' ')& nqtw z}
" $&(*0
5,.c$@3f@fg4!d!ddb`0\bx ~ppp@<4BdBd#(bg4+d+ddb`0\bppp@<4!d!d#(b:2___PPT9/0?%O5 )Potential Errors In Epidemiologic Studies**<*
zLearning Objectives"HB
Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB(Prevent confounding
Evaluate confounding")(6()ConfoundingHnA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outco
!"#$%&'()*+,./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefhijklmnopqrstuvwxyz{}~me and can even change the direction of the observed effect
0!0&v2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H( Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $DocumentSummaryInformation8Root EntrydO)XiCurrent User(SummaryInformation(g@PowerPoint Document(%
_%b
՜.+,0t
iPersonal Computerlu%ZjTimes New RomanTimes New Roman (Arabic)Blank Presentation*Potential Errors In Epidemiologic Studie0,2
kRestriction Strength H`&
=Straightforward
Convenient if criteria are narrow
Inexpensive>><> Restriction Limitation,<
BB4
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor,l Randomization B(
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization,m Randomization Strength(B4 Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor,$ Randomization Limitation(B4
.Not easy to perform
Ethical problems
Expensive//B/nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data,&oQ oMatching StrengthH&
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0f Matching Limitation(B4
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B<& Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor,Stratification LimitationB&
JDifficult if many variables with varying number of categories are requiredK!K,K}Multivariate Analysis<yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength<zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <!TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!,/67BCDEFHIJKLMNOPQRTUVWXY[
7B:P(
&
3vyd@) ?002<P`0P
l
CDv
H
0h ? r_zxb/(
fL/0DTimes New Roman(b(b4bj0XbXb`0DTimes New Roma
STP
&x~(aH
0h ? ̙33r5P.1b/(
fL/0DTimes New Roman(b(b4bj0XbXb`0DTimes New Roman (Arabic)bXb`0`
@n?" dd@ @@``H$.5"!
T "!
!nqtw z}" !#%(
+&'*,./2( $501c$@3f@fg4!d!ddb`0\bx ~ppp@<4BdBd#(bg4+d+ddb`0\bppp@<4!d!d#(b:2___PPT9/0?%O5)Potential Errors In Epidemiologic Studies**<*
zLearning Objectives"HB
Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB
(Prevent confounding
Evaluate confounding")(6(
)ConfoundingH
nA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outcome and can even change the direction of the observed effect
0!0 v2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H" Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $0,,
kRestriction Strength H`
=Straightforward
Convenient if criteria are narrow
Inexpensive>><
> Restriction Limitation,<
BB,
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor,
l Randomization B"
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization,
m Randomization Strength(B, Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor, Randomization Limitation(B,
.Not easy to perform
Ethical problems
Expensive//B
/nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data, oQ oMatching StrengthH
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0
f Matching Limitation(B,
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0
pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B< Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor,
Stratification LimitationB
JDifficult if many variables with varying number of categories are requiredK!K,
K}Multivariate Analysis<
yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength<
zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <
!TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!,
/67BCDEFHIJKLMNOPQRTUVWXY[rL1l(1g`b/(
fL/0DTimes New Roman(b(b4bj0XbXb`0DTimes New Roman (Arabic)bXb`0`
@n?" dd@ @@``H$.5"!
T "!
!nqtw z}" !#%(
+&'*,./2( $501c$@3f@fg4!d!ddb`0\bx ~ppp@<4BdBd#(bg4+d+ddb`0\bppp@<4!d!d#(b:2___PPT9/0?%O5)Potential Errors In Epidemiologic Studies**<*
zLearning Objectives"HB
Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB
(Prevent confounding
Evaluate confounding")(6(
)ConfoundingH
nA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outcome and can even change the direction of the observed effect
0!0 v2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H" Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $0,,
kRestriction Strength H`
=Straightforward
Convenient if criteria are narrow
Inexpensive>><
> Restriction Limitation,<
BB,
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor,
l Randomization B"
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization,
m Randomization Strength(B, Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor, Randomization Limitation(B,
.Not easy to perform
Ethical problems
Expensive//B
/nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data, oQ oMatching StrengthH
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0
f Matching Limitation(B,
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0
pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B< Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor,
Stratification LimitationB
JDifficult if many variables with varying number of categories are requiredK!K,
K}Multivariate Analysis<
yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength<
zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <
!TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!,
/67BCDEFHIJKLMNOPQRTUVWXY[INF( DHo
,
Cfyd@) ?002<P0Pp
r
S
H
0h ? ̙33r`lw`\b/(
fL/0DTimes New Roman(b(b4bj0XbXb`0DTimes New Roman (Arabic)bXb`0`
@n?" dd@ @@``H$.5"!
T "!
!nqtw z}" !#%(
+&'*,./2( $501c$@3f@fg4!d!ddb`0\bx ~ppp@<4BdBd#(bg4+d+ddb`0\bppp@<4!d!d#(b:2___PPT9/0?%O5)Potential Errors In Epidemiologic Studies**<*
zLearning Objectives"HB
Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB
(Prevent confounding
Evaluate confounding")(6(
)ConfoundingH
nA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outcome and can eve
!"#$%&'()*+,./0123456789:;<n change the direction of the observed effect
0!0 v2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H" Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $0,,
kRestriction Strength H`
=Straightforward
Convenient if criteria are narrow
Inexpensive>><
> Restriction Limitation,<
BB,
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor,
l Randomization B"
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization,
m Randomization Strength(B, Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor, Randomization Limitation(B,
.Not easy to perform
Ethical problems
Expensive//B
/nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data, oQ oMatching StrengthH
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0
f Matching Limitation(B,
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0
pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B< Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor,
Stratification LimitationB
JDifficult if many variables with varying number of categories are requiredK!K,
K}Multivariate Analysis<
yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength<
zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <
!TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!,
/67BCDEFHIJKLMNOPQRTUVWXY[
NNF ( (q@
,
Cvfyd@) ?002<P
r
S!v0`
H
0h ? ̙33rtYb/(
fL/0DTimes New Roman(b(b4bj0XbXb`0DTimes New Roman (Arabic)bXb`0`
@n?" dd@ @@``H$.5"!
T "!
!nqtw z}" !#%(
+&'*,./2( $501c$@3f@fg4!d!ddb`0\bx ~ppp@<4BdBd#(bg4+d+ddb`0\bppp@<4!d!d#(b:2___PPT9/0?%O5)Potential Errors In Epidemiologic Studies**<*
zLearning Objectives"HB
Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB
(Prevent confounding
Evaluate confounding")(6(
)ConfoundingH
nA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outcome and can even change the direction of the observed effect
0!0 v2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H" Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $0,,
kRestriction Strength H`
=Straightforward
Convenient if criteria are narrow
Inexpensive>><
> Restriction Limitation,<
BB,
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor,
l Randomization B"
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization,
m Randomization Strength(B, Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor, Randomization Limitation(B,
.Not easy to perform
Ethical problems
Expensive//B
/nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data, oQ oMatching StrengthH
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0
f Matching Limitation(B,
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0
pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B< Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor,
Stratification LimitationB
JDifficult if many variables with varying number of categories are requiredK!K,
K}Multivariate Analysis<
yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength<
zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <
!TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!,
/67BCDEFHIJKLMNOPQRTUVWXY[
YNF( z w˟
,
C vfyd@) ?002<P0
r
Sv
H
0h ? ̙33rvqVb/(
fL/0DTimes New Roman(b(b4bj0XbXb`0DTimes New Roman (Arabic)bXb`0`
@n?" dd@ @@``H$.5"!
T "!
!nqtw z}" !#%(
+&'*,./2( $501c$@3f@fg4!d!ddb`0\bx ~ppp@<4BdBd#(bg4+d+ddb`0\bppp@<4!d!d#(b:2___PPT9/0?%O5)Potential Errors In Epidemiologic Studies**<*
zLearning Objectives"HB
Understand the concept of confounding
Recognize the methods to prevent confounding
Know the methods to evaluate the impact of confounding0 {Performance Objectives"HB
(Prevent confounding
Evaluate confounding")(6(
)ConfoundingH
nA situation in which effects of two risk factors are mixed in the occurrence of the health problem under study&o!n6n
Confounding may lead to overestimation or underestimation of the true association between exposure and outcome and can even change the direction of the observed effect
0!0 v2 Criteria for Confounding<
Control of Confounding<
Prevention of Confounding<
j Restriction,
<H" Putting admissibility criteria for subjects and limiting enrollment into the study to individuals who fall within a specified category or categories of the confounder. $0,,
kRestriction Strength H`
=Straightforward
Convenient if criteria are narrow
Inexpensive>><
> Restriction Limitation,<
BB,
Reduces the number of subjects eligible to participate
Difficult if criteria are not narrow
Does not permit evaluation of association between exposure and outcome for varying levels of factor,
l Randomization B"
Every individual has the same chance of being classified in either of the two groups. The two methods commonly used for randomization are the use of a table of random numbers or the use of a computergenerated randomization,
m Randomization Strength(B, Controls confounders even those unsuspected
Study groups are comparable
Permits evaluation of association between exposure and outcome for varying levels of the factor, Randomization Limitation(B,
.Not easy to perform
Ethical problems
Expensive//B
/nMatching P Selecting study group and comparison group so that they are comparable with respect to various defined factors. In addition to the use of specific statistical tests for analyses of paired data, oQ oMatching StrengthH
eAppropriate when sample size is small and matching variables are limited
Study groups are comparable ff0
f Matching Limitation(B,
Difficult, expensive and time consuming to find comparison subjects
Matching on a particular variable prohibits studying its association with the outcome0
pEvaluation of ConfoundingB *Stratified analysis
Multivariate analysis+!+6* qStratified AnalysisH Stratification is a technique used to control confounding in the analysis stage that involves the evaluation of the association within homogeneous categories or strata of the confounding factor$,0 vStratification Strength*B< Easy for limited variables with limited number of categories
Permits evaluation of confounding and interaction
Permits evaluation of association between exposure and outcome for varying levels of the factor,
Stratification LimitationB
JDifficult if many variables with varying number of categories are requiredK!K,
K}Multivariate Analysis<
yAnalysis of data through construction of mathematical model that takes into account number of variables at the same timez!z0y Multivariate AnalysisStrength<
zDescribes efficiently the association between exposure and outcome taking in consideration the impact of other variables. {!{0z Multivariate AnalysisLimitation! <
!TThe choice of the appropriate model is complex and requires training and experienceU!U,T ~
Conclusion
H
There are number of methods for control of confounding. Each has its strength and limitation. In most situations, a combination of strategies provides better insights of data and more efficient control of confounding.!,
/67BCDEFHIJKLMNOPQRTUVWXY[rmn%b