Response Inhibition in AD/HD, CD, Comorbid AD/HD+CD, Anxious, and Control Children

15 Pages·1998·0.17 MB·English

J.ChildPsychol.Psychiat.Vol.39,No.3,pp.411–425.1998
CambridgeUniversityPress
’1998AssociationforChildPsychologyandPsychiatry
PrintedinGreatBritain.Allrightsreserved
0021-9630(cid:125)98$15.00(cid:173)0.00
(cid:125) (cid:125) (cid:173)
Response Inhibition in AD HD, CD, Comorbid AD HD CD,
Anxious, and Control Children: A Meta-analysis of Studies
with the Stop Task
Jaap Oosterlaan
UniversityofAmsterdam,TheNetherlands
Gordon D. Logan
UniversityofIllinois,Champaign,U.S.A.
Joseph A. Sergeant
UniversityofAmsterdam,TheNetherlands
Theaimofthisstudywastoinvestigatewhetherimpairedresponseinhibitionisuniquely
related to AD(cid:125)HD or whether deficits in response inhibition are also evident in other
psychopathological disorders. Furthermore, the suggestion was examined that anxiety
disorders are associated with abnormally high levels of response inhibition. This paper
presents the results of a meta-analysis of eight studies in which response inhibition was
assessed with the so-called stop task in five groups of children: children with attention
deficit(cid:125)hyperactivitydisorder(AD(cid:125)HD),childrenwithconductdisorder(CD),childrenwith
AD(cid:125)HD(cid:173)CD,childrenwithanxietydisorders,andcontrolchildren.Atotalof456children
participatedinthe8studies.Allchildrenwereintheagerange6–12years.Consistentand
robustevidencewasfoundforaresponseinhibitiondeficitinAD(cid:125)HD.However,response
inhibition deficits did not distinguish children with AD(cid:125)HD from children with CD, nor
fromchildrenwithcomorbidAD(cid:125)HD(cid:173)CD.Contrarytopredictions,anxiouschildrendid
notdemonstrateenhancedlevelsofresponseinhibition.
Keywords: Conduct disorder; anxiety disorder, attention deficit hyperactivity disorder,
informationprocessing,responseinhibition,meta-analysis.
Abbreviations:AD(cid:125)HD:attentiondeficit(cid:125)hyperactivitydisorder;BAS:BehaviouralActi-
vation System; BIS: Behavioural Inhibition System; CD: conduct disorder; IF: inhi-
bition function; LD: learning disorder; MRT: mean reaction time; NAS: Nonspecific
Arousal System; ODD: oppositional defiant disorder; SSRT: stop signal reaction time;
ZRFT-slope: slope of the IF plotted as a function of ZRFT (z score of the relative
finishingtime).
Introduction (1997) distinguishes three forms of response inhibition:
(1) inhibiting prepotent responses, (2) stopping an on-
Executivefunctionsarethosementalcontrolprocesses going response, and (3) inhibiting interference. The
that enable self-control and goal-directed behaviour ultimate goal of response inhibition is to enhance adap-
(Barkley,1996;Denckla,1994;Lezak,1983;Pennington tive functioning (Halperin, McKay, Matier, & Sharma,
& Ozonoff, 1996; Torgesen, 1994). These functions are 1994).
mediated by the frontal lobes, in particular by the Abnormalities in response inhibition are a central
prefrontalcortexanditsextendednetworks(Lezak,1983; component in the description and explanation of child
Pennington & Ozonoff, 1996; Torgesen, 1994). A fun- psychopathological disorders and, in particular, of at-
damental component of the executive functions is the tentiondeficit(cid:125)hyperactivitydisorder(AD(cid:125)HD).Barkley
ability to inhibit inappropriate responding. Barkley (1994,1997),forexample,arguedthatAD(cid:125)HDinvolves
a pervasive deficit in all forms of response inhibition.
According to Barkley, this deficit leads to secondary
impairments in four executive functions that depend on
Requests for reprints to: Dr Jaap Oosterlaan, Department
efficientresponseinhibitionfortheirexecution:working
of Clinical Psychology, University of Amsterdam, Roeters-
straat 15, 1018 WB Amsterdam, The Netherlands (E-mail: memory, internalisation of speech, self-regulation of
kp(cid:106)oosterlaan!macmail.psy.uva.nl). affect-motivation-arousal,andreconstitution.Asaresult,
411
412 J.OOSTERLAANetal.
children with AD(cid:125)HD present with a disorder charac- between, on the one hand, the process underlying
terised by impaired executive functions, which in turn responseexecution,andontheotherhand,theinhibitory
lead to disturbances in self-control and goal-directed process. This inhibitory process is triggered by infor-
behaviour. These disturbances ultimately result in the mation that tells the subject to discontinue or change
behavioursobservedinAD(cid:125)HD. a current course of action, such as an error during
Quay (1988a,b, 1997) used Gray’s (1987) neuro- performance. The process that finishes first determines
psychological model of brain function to explain the performance.Iftheinhibitoryprocessrunstocompletion
originofpoorresponseinhibitioninAD(cid:125)HD.InGray’s first, the response is inhibited. In the opposite case, the
model,behaviourisexplainedintermsoftheactivityof ongoing action is completed. In contrast to other mea-
twoopposingbrainsystems:the Behavioural Inhibition suresofresponseinhibition,thestoptaskenablesoneto
System(BIS),whichissensitivetosignalsofpunishment, investigate whether poor response inhibition is due to a
and the Behavioural Activation System (BAS), which is deficitintheinhibitoryprocess.
sensitive to signals of reward. The BIS serves to inhibit Briefly, the stop task requires fast and accurate
behaviour, whereas the BAS controls the initiation of executionofareactiontimetask,denotedastheprimary
behaviour. Quay proposed that children with AD(cid:125)HD task. Occasionally, a stop signal is presented, which
have a persistently underactive BIS, which results in requires the child to inhibit the response to the primary
responseinhibitiondeficits. task. Stop signals are presented at different intervals
Anumberofothertheoreticalaccountshaveemerged before the subject’s expected response. The shorter the
in which the inattentive, hyperactive, and impulsive interval, the more difficult it becomes to inhibit the
behaviour of children with AD(cid:125)HD has been suggested response.Usually,theintervalsarechosensuchthatthe
to arise from a deficit in response inhibition (Douglas, shortestintervalwillyieldaprobabilityofinhibitionclose
1988, 1989; Milich, Hartung, Martin, & Haigler, 1994; to 0, whereas the longest interval will produce a prob-
Newman&Wallace,1993;Pennington&Ozonoff,1996; abilityofinhibitioncloseto1.Ashortdescriptionofthe
Wender, 1972). Importantly, Barkley (1994, 1997) has stop task, the race model, and the main dependent
argued that response inhibition deficits are unique to measuresinthestoptaskisprovidedintheAppendix.
AD(cid:125)HD.Thissuggestionisalsomadeimplicitlyinother Inthelastfewyears,thestoptaskhasbeenutilisedin
theoretical accounts of AD(cid:125)HD (Douglas, 1988, 1989; a series of studies to investigate deficits in response in-
Pennington&Ozonoff,1996;Wender,1972). hibitioninchildrenwithAD(cid:125)HDandotherpsychopath-
ThepoorperformanceofAD(cid:125)HDchildrenonavariety ology (Aman, Roberts, & Pennington, in press;
of measures has been taken to support the response Daugherty, Quay, & Ramos, 1993; Jennings, Van der
inhibitiondeficithypothesisforAD(cid:125)HD.Thesemeasures Molen, Pelham, Brock, & Hoza, 1997; Oosterlaan &
include the Matching Familiar Figures Test (Campbell, Sergeant, 1996; Pliszka & Borcherding, 1995; Schachar
Douglas,&Morgenstern,1971;DuPaul,Anastopoulos, & Logan, 1990; Schachar & Tannock, 1995; Schachar,
Shelton, Guevremont, & Metevia, 1992; Weyandt & Tannock, Marriott, & Logan, 1995). This body of re-
Grant,1994),theContinuousPerformanceTask(seefor searchisreviewedinthecurrentstudy.Theprincipalaim
reviews,Barkley,Grodzinsky,&DuPaul,1992;Corkum of this study was to investigate whether data gathered
&Siegel,1993;seealsoHalperinetal.,1994),theGo(cid:125)No- withthestoptasksupportstheresponseinhibitiondeficit
goTask (Iaboni, Douglas, &Baker, 1995; Milich et al., hypothesis for AD(cid:125)HD. Furthermore, we investigated
1994; Shue & Douglas, 1992), delayed response tasks whetherimpairedresponseinhibitionisuniquelyrelated
(Daugherty & Quay, 1991; McClure & Gordon, 1984; toAD(cid:125)HDorwhetherdeficitsinresponseinhibitionare
Schweitzer & Sulzer-Azaroff, 1995; Solanto, 1990), the also evident in other psychopathological disorders.
Wisconsin Card Sorting Test (see for review, Barkley Finally, we examined the suggestion that anxiety dis-
etal.,1992),andmanyothers. orders are associated with abnormally high levels of
Thesemeasures,however,havebeencriticisedfortheir responseinhibition(Quay,1988a,b).
poorconstructvalidityandhavebeenconsideredastoo Meta-analytic procedures were used to aggregate
global (Halperin et al., 1994). Performance on these findingsacrossdifferentstudies.Oneoftheadvantagesof
measuresmay beinfluencedbymanyfactorsotherthan meta-analytic procedures is the greater power to detect
response inhibition, such as age and IQ (Milich et al., group differences (Rosenthal, 1991). This is particularly
1994; Schachar & Logan, 1990). The major criticism important given the small groups employed in most
levelled at these tasks is their failure to clarify the studies. Consequently, the failure to detect group dif-
mechanisms underlying impaired response inhibition ferences in these studies might reflect inadequate power
(Milich et al., 1994; Schachar & Logan, 1990). These ratherthantheabsenceofgroupdifferences.
criticismsdonotapplytothestoptask(Logan&Cowan, The existing literature enabled us to evaluate the
1984; Logan, Cowan, & Davis, 1984). This task is followingfivegroupcomparisonsthroughmeta-analysis.
purportedtomeasuretheabilitytointerruptanongoing Foreachofthesecomparisons,hypotheseswerederived
response. Several studies have supported the reliability fromcurrenttheoriesofchildpsychopathology.
and validity of the stop task as a measure of re- AD(cid:125)HD–controls (1). This comparison addresses
sponseinhibition(Kindlon,Mezzacappa,&Earls,1995; whetherdataderivedfromthestoptasklendsupportfor
Tannock,Schachar,Carr,Chajczyk,&Logan,1989). theoretical accounts of AD(cid:125)HD suggesting that this
Thestoptask isbased on a well-establishedtheory of disorder is characterised by poor response inhibition
response inhibition, known as the race model (see for (Barkley, 1994, 1997; Douglas, 1988, 1989; Newman &
review,Logan,1994;Logan&Cowan,1984).According Wallace, 1993; Pennington & Ozonoff, 1996; Quay,
to this model, response inhibition depends on a race 1988a,b, 1997; Wender, 1972). Thus, it was predicted
RESPONSEINHIBITIONINCHILDPSYCHOPATHOLOGY 413
thatAD(cid:125)HDchildrenwoulddemonstratepoorresponse haveshownthatastrongtendencyforresponseinhibition
inhibitioncomparedwithcontrols. in children is a powerful predictor of later anxiety
CD–controls (2) and AD(cid:125)HD–CD (3). As indicated disorders (see for review, Biederman, Rosenbaum,
earlier, it has been suggested that deficits in response Chaloff,&Kagan,1995).Thus,childrenwithanxietydis-
inhibition are confined to children with AD(cid:125)HD orderswerepredictedtodemonstrateenhancedresponse
(Barkley,1994,1997;Douglas,1988,1989;Pennington& inhibitionincomparisonwithcontrols.
Ozonoff,1996;Wender,1972).However,consonantwith
AD(cid:125)HD, poor response inhibition has also been de-
Method
scribedasacorephenomenoninconductdisorder(CD)
(Farrington, 1993; Milich et al., 1994; Newman & Description of the Studies
Wallace, 1993; Quay, 1988a,b, 1993). Thus, response
inhibition deficits may not be specific to AD(cid:125)HD, i.e. Thisreviewcoversstudiesconductedbetween1990and1997.
Eight independent published and unpublished studies were
impairedresponseinhibitionmaynotbethehallmarkof
identifiedthatmetthefollowinginclusioncriteria:(1)thestudy
AD(cid:125)HD.
contained one or more of the psychopathological groups of
According to Quay (1988a,b, 1993, 1997), both interest, and (2) the study included a comparison group of
AD(cid:125)HDandCDareassociatedwithresponseinhibition controlchildren.Allstudiesfocusedonchildrenintheagerange
deficits. However, the dysfunction underlying poor re- 6to12years.Abriefdescriptionofstudyattributesisprovided
sponse inhibition differs for the two disorders. Quay in Table 1. In some reports insufficient information was pro-
(1988a,b, 1993) has suggested that CD reflects an over- vided to complete the meta-analysis. In these cases, authors
active BAS that dominates the BIS. According to this werecontactedtoobtainthemissinginformation(cid:34).
Seven studies contained a group of AD(cid:125)HD children.
view,theexcessiveBASactivitycausesastrongtendency
Schacharet al. (1995) reportedtheresults for threegroups of
torespondandinterfereswiththecapabilityforresponse
AD(cid:125)HDchildren:apervasivegroup,aschool-onlygroup,and
inhibition. By contrast, poor response inhibition in
ahome-onlygroup.Onlythepervasivegroupwasincludedin
AD(cid:125)HDresultsfromanunderactiveBIS(Quay,1988a,b,
the meta-analysis to avoid the problem of correlated results
1997). (Rosenthal,1991)andtomaximisethechanceoffindinggroup
Quay referred in particular to children with under- differences (Luk, Leung, & Yuen, 1991; Schachar & Logan,
socialised aggressive conduct disorder (American Psy- 1990; Schachar et al., 1995; Van der Meere, Wekking, &
chiatricAssociation,1987),agroupofchildrencurrently Sergeant,1991).
alluded to as CD–childhood-onset type (American Psy- Children with oppositional defiant disorder (ODD) or CD
chiatric Association, 1994). This type of CD generally were included in four of the studies. Five studies included a
comorbidAD(cid:125)HD(cid:173)CDgroup.Finally,threestudiesincluded
emerges prior to age 10 years. Given the age range of
children with anxiety disorders. Children with ODD or CD
childrenwithCDinthepresentmeta-analysis(6–12years
wereregardedasasinglegroupandfurtherdealtwithunderthe
old), Quay’s hypothesis concerning CD is particularly
headingCD.ODDisfrequentlyseenasamilderformofCD,
relevant.
thetwodisordersarerelatedtothesameriskfactorsandforms
Theexistingdataderivedwiththestoptaskenabledus ofimpairment,andODDtosomedegreepredictstheonsetof
to investigate both differences between CD and control CD (Achenbach, 1993; American Psychiatric Association,
childrenandbetweenAD(cid:125)HDandCDchildren.Theaim 1991; Lahey, Loeber, Frick, Quay, & Grimm, 1992; Loeber,
wastodeterminewhetherresponseinhibitiondeficitsare Green, Kennan, & Lahey, 1995). A total of 456 children
uniquetochildrenwithAD(cid:125)HDorwhetherthesedeficits participatedinthe8studies.
arealsoevidentinchildrenwithCD.Thestoptaskwould
beparticularlyusefulinclarifyingthenatureofAD(cid:125)HD,
Dependent Variables
ifitisabletodistinguishAD(cid:125)HDchildrenfromchildren
withCD. Themeta-analysisfocusedonthefollowingfourdependent
AD(cid:125)HD(cid:173)CD–AD(cid:125)HD (4). This comparison was variablesderivedfromthestoptask(adetaileddescriptionof
thesemeasuresisprovidedintheAppendix):
designed to investigate whether children with comorbid
AD(cid:125)HD(cid:173)CDdifferfromchildrenwithAD(cid:125)HDonly.It (1) Meanreactiontime(MRT),whichmeasuresthelatency
hasbeensuggestedthatthecombinationofAD(cid:125)HDand oftheprocessesinvolvedinresponseexecution.Sincethis
CD represents a distinct nosological entity (Biederman, meta-analysis focused on abnormalities in response
Newcorn, & Sprich, 1991; Schachar & Tannock, 1995). inhibition, MRT was not the main focus of the present
Following Quay’s (1988a,b, 1993, 1997) model, one study.
might speculate that the AD(cid:125)HD(cid:173)CD group shows (2) Theinhibitionfunction(IF),whichreflectstheefficiency
of the inhibitory mechanism controlling for differences
greater deficiencies in the capability for response in-
in MRT. Most researchers take the slope of this IF
hibitionduetoacombinationofunderactivityintheBIS
(IF-slope) as an index of the subject’s capability for
(resultinginpoorresponseinhibition)andoveractivityin
responseinhibition.TheflattertheIF-slope,thepoorer
the BAS (which may cause an exaggerated tendency to thecapabilityforresponseinhibition(Logan,1994).
respond). Thus, response inhibition deficits in children According to the race model (see for review, Logan,
with AD(cid:125)HD(cid:173)CD might be more pronounced than in 1994; Logan & Cowan, 1984), the IF is determined by
childrenwithAD(cid:125)HDalone.
Anxietydisorder–controls(5). IncontrasttoAD(cid:125)HD (cid:34)For two studies data on IF-slope and ZRFT-slope were not
and CD, anxiety disorders have been associated with
included in the meta-analysis. Jennings et al. (1997) did not
abnormallyhighlevelsofresponseinhibition.Ithasbeen
calculate within-group standard deviations for these two
suggested by Quay (1988a,b) that an overactive BIS dependent measures. Part of the data from the Schachar and
underliesanxietydisorders.Furthermore,severalstudies Tannock(1995)studywerenotavailable.
414 J.OOSTERLAANetal.
Table1
StudiesIncludedinMeta-analysis:SubjectandTaskCharacteristics
Study Subjects Agea Subjectselection Taskcharacteristics
1. Schachar& 10Controls 10.0 AD(cid:125)HDdiagnosisbasedonparentinterview Stoptask
Logan(1990)b 13AD(cid:125)HD 9.3 and(cid:125)orteacherratings.Otherdiagnoses Primarytask:two-choice
9CD(cid:125)ODD 9.8 basedonparentinterviewalone.Control reactiontimetask(lettersX
14AD(cid:125)HD(cid:173)CD(cid:125)ODD 9.3 childrenwerefreeofanydiagnosis. andO)
13Emotionaldisorderc 9.9 DiagnosesaccordingtoDSMIII(-R) 25%stoptrials,stopsignals
11LDd 10.0 criteria.CriteriaforLDdiagnosiswereat 500,400,300,200,100,
leastaverageIQandlowreadingattainment. 0msecbeforeMRT
AllchildrenhadIQs(cid:38)80andwerefreeof 432trials(21.6min),2breaks
medication. (cid:173)96practicetrials
2. Daugherty, 15Controls 11.0 Groupassignmentbasedonteacherratings. SeeStudy1
Quay,& 9AD(cid:125)HD 11.4 Psychopathologicalgroupshadelevated
Ramos(1993) 8CD 11.1 scoresonrelevantscaleandlowscoreson
11CD(cid:173)AD(cid:125)HD 11.7 otherscales.Controlchildrenhadlowscores
12Anxiety-withdrawal 10.8 onallscales.
3. Aman, 22Controls 12.1 Assignmenttooneofthegroupsbasedon Stoptask
Roberts,& 22AD(cid:125)HDe 12.1 parentstructuredinterviewandparent Primarytask:seeStudy1
Pennington questionnaires.AD(cid:125)HDchildrenshowed 1(cid:125)3stopsignaltrials,stop
(inpress) favourableresponsetostimulants.Diagnoses signals500,350,250,
accordingtoDSMIII-Rcriteria.Children 100msecbeforeMRT
withotherpsychiatricdisordersandreading 192trials(9.6min)
disabilitywereexcluded.AllchildrenhadIQs (cid:173)96practicetrials
(cid:38)80andwerefreeofmedicationf.
4. Pliszka& 31Controls 8.8 AD(cid:125)HDdiagnosesbasedonparentstructured Stoptask
Borcherding 26AD(cid:125)HD(or(cid:173)ODD) 8.7 interviewandteacherratings.CDdiagnoses Primarytask:two-choice
(1995) 8AD(cid:125)HD(cid:173)CD 8.2 basedonparentstructuredinterviewalone. reactiontimetask(red
17AD(cid:125)HD(cid:173)overanxious 9.4 Diagnosesofoveranxiousdisorderbasedon andgreenlight)
(or(cid:173)ODD)d childstructuredinterview.Absenceof 25%stoptrials,stopsignals
18psychiatric 9.5 psychiatricdisorderincontrolswasassessed 500,400,300,200,100,
controlsd withavarietyofmeasures. 0msecbeforeMRT
432trials(21.6min)
(cid:173)48practicetrials
5. Schachar& 16Controls 9.0 Diagnosesbasedonparentand(cid:125)orteacher SeeStudy6
Tannock 22AD(cid:125)HD 9.2 structuredinterview.Controlchildrenwere
(1995)g 5CD 10.1 freeofanydiagnosisandlearningproblems.
18AD(cid:125)HD(cid:173)CD 8.8 DiagnosesaccordingtoDSMIII-Rcriteria.
AllchildrenhadIQs(cid:38)80and(cid:37)130and
werefreeofmedication.
6. Schachar, 22Controls 9.2 Diagnosesbasedonparentand(cid:125)or Changetask
Tannock, 10Home(cid:175)onlyAD(cid:125)HDd 9.4 (dependingonAD(cid:125)HDsubtype)teacher Primarytask:seeStudy1
Marriott,& 9School(cid:175)onlyAD(cid:125)HDd 9.8 structuredinterview.Controlchildrenwere 25%stopsignalstrials,stop
Logan(1995)h 14PervasiveAD(cid:125)HD 8.7 freeofanydiagnosisandLD.AD(cid:125)HD signals500,350,200,
childrenhadnoadditionaldiagnosisofODD 50msecbeforeMRT
orCD.DiagnosesaccordingtoDSMIII-R 288trials(14.4min),2breaks
criteria.AllchildrenhadIQs(cid:38)80andwere (cid:173)aminimumof72practice
freeofmedication. trials
7. Oosterlaan& 17Controls 8.7 Assignmenttooneofthepsychopathological Stoptask
Sergeant 15AD(cid:125)HDi 9.3 groupsbasedonparent,teacherandchild Primarytask:spatial
(1996) 18ODD(cid:125)CD 9.3 ratings(agreementrequiredbetweentwo compatibletwo-choice
20Anxietydisorder 10.1 informants).Controlchildrenobtainedlow reactiontimetask(white
scoresonallscalesofallquestionnaires.All squares)
childrenhadIQs(cid:38)80andwerefreeof 25%stoptrials,stopsignals
medication. 500,350,200,
50msecbeforeMRT
256trials(12.8min),1break
(cid:173)64practicetrials
RESPONSEINHIBITIONINCHILDPSYCHOPATHOLOGY 415
Table1(cont.)
Study Subjects Agea Subjectselection Taskcharacteristics
8. Jennings,Van 26Controls 9.8 AD(cid:125)HDdiagnosisbasedonparentstructured Stoptask
derMolen, 40AD(cid:125)HDd,j 9.7 interviewandquestionnairescompletedby Primarytask:simplereaction
Pelham, 25AD(cid:125)HD(cid:173)ODD(cid:125)CDj 9.1 parentandteacher.Diagnosesaccordingto timetaskembeddedinvideo
Brock,& DSMIII-Rcriteria.Controlchildren gameformat(stoplight
Hoza obtainedlowscoresonparentandteacher changingfromredtogreen)
(1997) questionnaires.Allchildrenwerefreeof 30%stoptrials,stopsignals
medication. 100or200msecaftergo
stimulus
Variabletriallength,
presentationoftaskstimuli
relativetocardiaccycleand
respiratoryphase
200trials(40min),1break
(cid:173)aminimumof25practice
trials
aMeanageinyears.
bPathologicalgroupscouldhaveadditionaldiagnosisofemotionaldisorderorLD.
cThetermemotionaldisorderencompassesmainlyanxietydisorders.
dGroupsnotincludedinmeta-analysis.
eAD(cid:125)HDgroupobtainedhighratingsofaggressiveanddelinquentbehaviourononeoftheparentquestionnaires.
fSubjectsweretestedtwice,oneweekapart.AD(cid:125)HDchildrenwereonmethylphenidateinthefirstsessionandunmedicatedinthe
secondsession.Theresultsofthesecondsessionwereenteredintothemeta-analysis.
gLowerreadingattainmentinAD(cid:125)HDandAD(cid:125)HD(cid:173)CDgroupsthaninCDandcontrolgroup.Lowerarithmeticattainmentin
CDandAD(cid:125)HD(cid:173)CDgroupsthaninAD(cid:125)HDandcontrolgroup.
hFiveAD(cid:125)HDchildrenmetcriteriaforLD,twochildrenmetcriteriaforoveranxiousdisorder.
iSixchildrenwithAD(cid:125)HDalsometcriteriaforinclusioninODD(cid:125)CDgroup.
jTwenty-five of the 40 AD(cid:125)HD children carried a concurrent ODD diagnosis and three children met criteria for CD. Jennings
etal.reportedboththeresultsforthefullsampleandforthesubsampleofchildrenwithcomorbiddisorders.Threechildrenwere
dropped from this subsample due to incomplete data. Only data of the subsample were included in the meta-analysis. For this
subsample,Jenningsetal.didnotspecifythenumberofchildrenmeetingcriteriaforODDandforCD.AD(cid:125)HDgrouphadlower
reading,spelling,andarithmeticattainmentthancontrolgroup.NineAD(cid:125)HDchildrenmetcriteriaforLD.
parameters of both the response execution process and (1989).Calculationswereweightedbysamplesize.Theresults
the inhibitory process. Using the race model, two wereevaluatedacrossstudiesintermsoftheireffectsizeandin
measurescanbederivedtoexaminewhetheradeficiency termsoftheirsignificancelevels.Furthermore,diffusetestswere
intheinhibitoryprocessunderliesapoorIF:stopsignal conducted to assess the consistency across studies of both
reactiontime,andtheslopeoftheIFplottedasafunction estimatedeffectsizesandsignificancelevels.
ofZRFT.
(3) Stopsignalreactiontime(SSRT),whichisanestimateof
thelatencyoftheinhibitoryprocess.SSRTisoneofthe Possible Moderating Variables
parametersthatdeterminestheprobabilityofinhibition
Before presenting the results of the meta-analysis, several
givenastopsignal.Theslowertheinhibitoryprocess,the
pertinent study characteristics are described. Each of these
harderitbecomestoinhibittheresponsetotheprimary
studycharacteristicsisevaluatedforitspotentialimpactonthe
task(Logan,1994).
findingsobtained.
(4) The slope of the IF plotted as a function of ZRFT
Subjectselection. Studiesdifferedsharplyinthecriteriaand
(ZRFT-slope).InadditiontoaslowSSRT,poorresponse
the measures that were used to select the various groups (see
inhibitioncouldreflecttwootherdeficitsintheinhibitory
Table1).Thisprobablyexplainspartoftheheterogeneityinthe
process.First,itmightindicatethattheinhibitoryprocess
resultsacrossstudies.
wastriggeredlessoften.Second,poorresponseinhibition
couldreflectgreatervariabilityinlatencyoftheinhibitory In five studies diagnoses were made using the third revised
process (Logan, 1994). These two parameters of the edition of the Diagnostic and statistical manual of mental dis-
inhibitory control process are reflected in the ZRFT- orders(DSM III-R; AmericanPsychiatric Association, 1987).
slope. With the exception of the Daugherty et al. (1993) study,
pathological groups were selected from clinic referred or
Thefollowing informationwas obtainedfor all studies and otherwise treated children (e.g. placed in special educational
served as the basis of the meta-analysis: (1) the number of facilities).
subjectsineachgroup,andforeachofthedependentmeasures, The presence or absence of psychopathological symptoms
(2) the group mean, and (3) the within-group standard was assessed by means of reports from parents, teachers,
deviation. From these data, effect size estimations and their children,oranycombinationoftheseinformants.Avarietyof
direction were calculated for all relevant group comparisons measures was used across the studies, including standardised
withineachofthestudies.Theresultsoftheseanalyses,inturn, ratingscales,structureddiagnosticinterviews,oracombination
were subjected to meta-analytic procedures (Mullen, 1989; ofboth.Insomestudies,diagnoseswerebasedonthereportof
Rosenthal, 1991). All analyses were conducted with the ad- a single source. For example, in the Daugherty et al. (1993)
vancedBASICmeta-analysisprogrammedevelopedbyMullen study,teacherswerethesoleinformants.Inotherstudies,twoor
416 J.OOSTERLAANetal.
three informants were used. In the majority of these studies, performanceonthestoptask(Daughertyetal.,1993;Pliszka&
the reports by the different sources were not necessarily re- Borcherding,1995).Innoneofthestudieswereagedifferences
quired to converge. Since the degree of consistency between noted between the groups. This suggests that age is not a
different informants’ reports is typically modest (Achenbach, confounding variable in any of the studies. The mean age of
McConaughy,&Howell,1987),theuseofmultipleinformants control children ranged from 8.7 years in the Oosterlaan and
results in a greater chance of detecting children with actual Sergeant(1996)studyto12.1yearsintheAmanetal.(inpress)
disorders. study.
Other studies used a more restrictive approach to combine Intellectual functioning. IQ was assessed in six studies. In
the information from different informants. In four studies, five of these studies, children with below-average intellectual
children received a diagnosis of AD(cid:125)HD if both parent and functioning(IQlessthan80)wereexcluded.Theremainingtwo
teacherreportsindicateditspresence.OosterlaanandSergeant studies did not provide information on the children’s IQ
(1996)usedparentandteacherratingsandaself-reportanxiety (Daugherty et al., 1993; Pliszka & Borcherding, 1995). Since
questionnaire to select AD(cid:125)HD, CD, anxiety disorder and several studies have demonstrated that IQ is not related to
controlchildren.Childrenwereassignedtooneofthepsycho- children’sperformanceonthestoptask(Kindlonetal.,1995;
pathological groups only if the reports of two informants Oosterlaan & Sergeant, 1996; Schachar & Logan, 1990;
convergedonthisclassification. Schacharetal.,1995),itseemsunlikelythatgroupdifferencesin
Thedistinction between pervasive and situational disorders intellectualfunctioningunderlieanyofthefindings.
seems important, at least for AD(cid:125)HD. Several studies have Medication. Insixstudies,childrenwerefreeofmedication
indicated that children with pervasive AD(cid:125)HD show greater or discontinued medication before being tested with the stop
cognitivedeficitsthanchildrenwithsituationalspecificAD(cid:125)HD task. Daugherty et al. (1993) and Pliszka and Borcherding
symptoms (e.g. Luk et al., 1991; Schachar & Logan, 1990; (1995)didnotreportonthepossibleuseofmedication.
Schacharetal.,1995;VanderMeereetal.,1991). Format of the stop task. In six studies the stop task was
Comorbidity. A large body of research has demonstrated used, whereas two studies (Schachar & Tannock, 1995;
that AD(cid:125)HD shows considerable overlap with a number of Schacharetal.,1995)utilisedamodificationofthestop task,
otherdisorders,suchasODDandCD,mooddisorders,anxiety known as the change task (De Jong, Coles, & Logan, 1995;
disorders,and learningdisabilities(Barkley,1990; Biederman Logan & Burkell, 1986). The stimuli in the change task are
etal.,1991;Semrud-Clikemanetal.,1992).Comorbiditymight identicaltothoseinthestoptask.However,thetwotasksdiffer
haveinfluencedchildren’sperformanceonthestoptask. withrespecttothedemandexertedbythestopsignal.Inboth
In all but two of thestudies (Oosterlaan & Sergeant, 1996; tasks,thestopsignalinstructssubjectstoinhibittheirresponse
Pliszka & Borcherding, 1995), AD(cid:125)HD children were free of to the primary task. In the change task, the stop signal, in
comorbidODDorCD.IntheOosterlaanandSergeantstudy,6 addition,requiressubjectstoimmediatelyre-engageinanother
out of the 15 children showed associated ‘‘aggressive or response,i.e.toexecutetheso-calledchangeresponse.Several
delinquent’’symptoms.TheAD(cid:125)HDsamplestudiedbyPliszka studies have indicated that the response execution and in-
and Borcherding contained children with AD(cid:125)HD alone and hibitoryprocessesseemtofunctionessentiallythesameinboth
AD(cid:125)HD with an additional ODD diagnosis. However, the the stop task and the change task. However, the change task
investigatorsdidnotspecifythecorrespondingproportions. appearstoexerthighercognitiveprocessingdemandsthanthe
In four studies, children were examined for learning dis- stoptask,sinceitcarriestheadditionaldemandofre-engaging
abilities,althoughinmostofthesestudiestheassessmentwas in another response (De Jong et al., 1995; Logan & Burkell,
limited to reading disabilities (Amanet al.,in press; Jennings 1986).
etal.,1997;Schachar&Logan,1990;Schacharetal.,1995).In In five studies an identical two-choice reaction time task
only one study were children with concurrent reading dis- servedastheprimarytask.Inthistask,childrenrespondedto
abilities excluded (Aman et al., in press). The remaining four thelettersXand Obypressingone ofthetwo corresponding
studies did not report on the possible presence of learning buttonsonaresponsebox.Childrenrespondedwithaleft-and
disabilities. aright-handfinger,usuallytheirindexfingers.Inthosestudies
Recent findings highlight the necessity to control for asso- inwhichthechangetaskwasused,childrenrespondedwithtwo
ciated learning disabilities. A recent study by Tannock and separatefingersofthelefthand.Modificationsofthisprimary
Marriott (1992) suggested that learning disabilities, and not taskwereusedintwootherstudies.InthestudybyPliszkaand
AD(cid:125)HD,wereassociatedwithpoorresponseinhibitionanda Borcherding(1995),childrenrespondedtoaredandgreenlight.
slowerinhibitoryprocess.Furthermore,Daughertyetal.(1993) In the Oosterlaan and Sergeant (1996) study, a spatial com-
havereportedamoderatepositivecorrelationbetweenreading patibletwo-choicereactiontimetaskservedastheprimarytask.
achievementandIF-slope. Jennings et al. (1977) used a simple reaction time task
Conflicting predictions were made regarding the capability embeddedinavideogameformat.Inthistask,agreentraffic
for response inhibition in children with anxiety disorders and light signalled children to movea mouse-controlled ice-cream
in children with externalising disorders (AD(cid:125)HD, CD, and carttoatargetareaonthecomputerscreen.
AD(cid:125)HD(cid:173)CD). Consequently, the distinction between these With the exception of the study by Jennings et al. (1997),
two disorders seems important. In four of the eight studies, thetiming of stimuli was similar for all studies. Trials started
children with externalising disorders were free of anxiety with the presentation of a warning signal of 500msec. Im-
disorders (Aman et al., in press; Daugherty et al., 1993; mediately thereafter the primary task stimulus was displayed
Oosterlaan&Sergeant,1996;Pliszka&Borcherding,1995).In for1000msec.Anintertrialintervalof1500msecwasused.In
theSchacharandLogan(1990)study,anadditionaldiagnosisof allstudiesprimarytaskstimuliwerepresentedvisually.Jennings
anxietydisorderwasallowedinallpsychopathologicalgroups. etal.investigatedtherelationshipbetweenresponseinhibition
However,exactfigureswerenotpresented.TheAD(cid:125)HDsample and heartbeat timing. In this study, stimuli were presented
ofSchacharetal.(1995)includedtwochildrenwithconcurrent relativetothechild’scardiaccycleandrespiratoryphase.Note
overanxious disorder. In the remaining two studies, children thatthepresentationrateofstimuliininformationprocessing
werenotassessedforthepossiblepresenceofassociatedanxiety taskshasbeenshowntoplayacrucialroleintheperformance
disorders(Jenningsetal.,1997;Schachar&Tannock,1995). ofAD(cid:125)HDchildren(seeVanderMeere,1996,forreview).
Gender and age. In five studies, boys served as subjects, Jennings et al. (1997) presented stop signals relative to the
whereasintheremainingthreestudiesbothsexeswereincluded. onsetoftheprimarytaskstimulus,whereasintheotherstudies
Gender, however, does not seem to influence the child’s stop signals were presented relative to the child’s MRT (see
RESPONSEINHIBITIONINCHILDPSYCHOPATHOLOGY 417
Appendix). In all studies, the longest stop signal interval was number of subjects across studies for the contrasting
500msec.Theshorteststopsignalintervalvariedfrom0msec groups, (b) the number of relevant studies, (c) the
to100msec.Insixofthestudies,stopsignalswerepresentedon weighted mean for both groups, (d) the combined
25%ofthetrials.InthestudiesbyJenningsetal.(1997)and estimatedeffectsizeintermsofCohen’sd(1988),and(e)
Amanetal.(1995),thepercentageofstoptrialswassomewhat
themeta-analyticZwiththerespectivesignificancelevel.
higher:30%and33.3%,respectively.
InagreementwithCohen’sguidelines,effectsizesof.20,
Stopsignalswerepresentedintheauditorymodality.Inthe
.50, and .80 were used as thresholds to define small,
Jennings et al. (1997) study, the stop signal resembled the
medium,andlargeeffects,respectively.
soundofacarhorn.Intheotherstudies,atonewaspresented
bythecomputerorthroughearphones(Oosterlaan&Sergeant, MRT. Meta-analytic results for the latency of res-
1996).Wherereported,toneswere1kHzinpitchand100msec pondingarepresentedinTable2.MRTsofAD(cid:125)HDand
induration. control children were compared in seven studies. The
Studies with young adults have demonstrated that the stop meaneffectsize(d(cid:175).49)wasclosetoCohen’sstandard
task is reasonably robust for differences in the format of the of .50 for a medium effect, indicating 32% nonoverlap
task. First, a number of studies have shown that measures between the MRT distributions of the two groups.
derivedfromthestoptaskarenotaffectedbydifferencesinthe
Differences between the other four pairs of groups were
natureoftheprimarytask.Findingswiththestoptaskdonot
small and not significant. Note that the results of all
seemtovaryasafunctionoftaskdifficulty,noraretheresults
group comparisons were consistent across studies, both
differentfortasksthatrequirecontinuousordiscreteresponses
in terms of effect sizes and significance levels. This
(seeforreviews,Logan,1994;Logan&Cowan,1984).Second,
thestoptaskhasbeenfoundrobustforchangesinstopsignal indicatesthattheresultsofthestudiesagreebothonthe
probability. Typically, the primary task stimulus occurs on magnitudeofthegroupdifferencesandonthesignificance
every trial and the stop signal occurs on 25% of the trials. ofthesedifferences.
Logan(1981)andLoganandBurkell(1986)studiedtheeffectof NotealsothatacomparisonofweightedmeanMRTs
varying stop signal probability from 10% to 80%. Measures of AD(cid:125)HD and CD children suggests slower MRTs for
derivedfromthestoptaskwerenotmuchaffectedwhenthestop AD(cid:125)HD children. The combined effect size and signifi-
signal probability varied between 10% and 50%. Third, and
cance level, however, indicated that the two groups did
finally,Logan(1994;Logan&Cowan,1984)reviewedstudies
not differ in MRT. This finding could be explained as
using auditory and(cid:125)or visual stop signals and concluded that
follows.Acrossstudies,substantialdifferenceswerenoted
performance on the stop task is not affected by stop signal
inMRTs.Forexample,MRTsofcontrolchildrenvaried
modality. In summary, the stop task has been found to be
reasonablyrobustfor differencesinthenature oftheprimary fromabout350msec(e.g.Oosterlaan&Sergeant,1996)
task, for differences in stop signal probability, and for dif- to 900msec (e.g. Schachar & Logan, 1990). These
ferences in stop signal modality. Therefore, it seems unlikely differences reflect variations in the cognitive load of the
thatdifferencesintheformatofthestoptaskunderlieanyofthe tasks employed. Since a greater proportion of CD than
findings. AD(cid:125)HD children was tested with a stop task with
Between-studies differences were noted in the nature and relativelylowcognitivedemands(asindexedbytheMRT
amount of practice before starting the task. Furthermore,
ofthecontrolgroup),CDchildrenshowedasubstantially
studiesvariedinthenumberoftrialsandthenumberofbreaks shorterweightedmeanMRTthanAD(cid:125)HDchildren.
withinthetask.Estimatesoftime-on-taskrangedfrom9.6min
IF-slope. Table 3 displays the results for IF-slope.
to 21.6min (since in the majority of studies the length of the
Acrosssixstudies,AD(cid:125)HDchildrenhadflatterIF-slopes
taskwasnotspecified,anestimateoftime-on-taskwasobtained
bymultiplyingthenumberoftrialswiththetriallength).Inthe than controls, suggesting impairments in response in-
Jenningsetal.(1997)study,thelengthoftrialsvaried.Inthis hibition. The combined effect size of d(cid:175).94 exceeds
study,fourblockswereadministered,each10mininduration. Cohen’s threshold for a large effect and corresponds to
Details on the instructions given to the children varied 53% nonoverlap between the IF-slope distributions of
considerably between studies. Instructions are considered im- the two groups. Diffuse tests indicated that effect sizes
portant in the stop task, since the task involves a delicate were consistent across studies, whereas heterogeneity
balancebetweenfastandaccurateresponding,ontheonehand,
was noted for the significance levels. This heterogeneity
andinhibitionofresponses,ontheotherhand.Toenhancethe
in significance levels was attributable mainly to the
child’s performance, Jennings et al. (1997) used feedback, Daugherty et al. (1993) study, in which AD(cid:125)HD and
monetary reward, and monetary loss. Where reported, the
control children showed highly comparable IF-slopes.
experimenterremainedwiththechildduringthetask.However,
noneofthestudiesprovidedinformationabouttheroleofthe In all other studies, AD(cid:125)HD children showed signifi-
experimenter during the task. Most studies failed to report cantlyflatterIF-slopesthancontrols.
whether the experimenter was aware of the child’s group Similar findings were obtained for the comparison
assignment and the purpose of the study. Where reported, betweenCDandcontrolchildren.Averagedacrossthree
subjectsweretestedindividually. studies, IF-slopes were flatter in CD children than in
controls. A medium combined effect size of d(cid:175).56 was
found, which translates into 36% nonoverlap between
Results
theIF-slopedistributionsofthetwogroups.Resultswere
As indicated previously, this meta-analysis concen- uniform across studies, both in terms of effect sizes and
tratesonfivegroupcomparisons,namely:(1)AD(cid:125)HD– significancelevels.
controls, (2) CD–controls, (3) AD(cid:125)HD–CD, (4) In three studies AD(cid:125)HD and CD children were
AD(cid:125)HD(cid:173)CD–AD(cid:125)HD, and (5) anxiety disorder– compared. Although AD(cid:125)HD children had somewhat
controls.Meta-analyticresultsforeachofthedependent flatter IF-slopes than children with CD, this difference
measuresarepresentedinTables2through5.Thetables was small and not significant. Thus, no support was
show, for each of the five comparisons: (a) the total obtainedforthehypothesisthatAD(cid:125)HDchildrenwould
418 J.OOSTERLAANetal.
Table2
GoResponseMeanReactionTime(MRT;inMsec)
Samplesize Weightedmean
Groupcomparison Group1 Group2 Studies Group1 Group2 d CombinedZ
AD(cid:125)HD–Controls 121 133 7 772.6 702.5 0.49 3.67*
CD–Controls 40 58 4 612.6 642.0 0.29 1.59
AD(cid:125)HD–CD 59 40 4 726.1 612.8 0.36 1.73
AD(cid:125)HD(cid:173)CD–AD(cid:125)HD 51 70 4 867.2 847.5 0.16 0.87
Anxious–Controls 45 42 3 623.0 607.7 0.25 1.23
Diffusecomparisonswereconductedtoassessthehomogeneityofeffectsizesandsignificance
levels.Allgroupcomparisonsyieldedhomogeneousresults.
*p(cid:33).001.
Table3
InhibitionFunctionSlope(IF-slope)
Samplesize Weightedmean
Groupcomparison Group1 Group2 Studies Group1 Group2 d CombinedZ
AD(cid:125)HD–Controls 99 117 6 10.2 14.8 0.94a 6.36**b
CD–Controls 35 42 3 12.6 16.0 0.56a 2.35*a
AD(cid:125)HD–CD 37 35 3 11.5 12.8 0.25a 0.73a
AD(cid:125)HD(cid:173)CD–AD(cid:125)HD 33 48 3 10.0 8.6 0.17b 0.73b
Anxious–Controls 45 42 3 14.4 15.9 0.34a 1.41a
Diffusecomparisonswereconductedtoassessthehomogeneityofeffectsizesandsignificance
levels:aHomogeneouseffect;bHeterogeneouseffect.
*p(cid:33).05;**p(cid:33).001.
Table4
StopSignalReactionTime(SSRT;inMsec)
Samplesize Weightedmean
Groupcomparison Group1 Group2 Studies Group1 Group2 d CombinedZ
AD(cid:125)HD–Controls 121 133 7 349.4 246.4 0.64a 4.97**a
CD–Controls 40 58 4 265.7 248.0 0.51b 2.64*b
AD(cid:125)HD–CD 40 59 4 364.7 265.7 0.07b 0.37b
AD(cid:125)HD(cid:173)CD–AD(cid:125)HD 51 70 4 323.6 361.8 0.13a 0.78a
Anxious–Controls 45 42 3 231.3 207.6 0.20a 0.88a
Diffusecomparisonswereconductedtoassessthehomogeneityofeffectsizesandsignificance
levels:aHomogeneouseffect;bHeterogeneouseffect.
*p(cid:33).01;**p(cid:33).0001.
be more deficient in response inhibition than children SSRT. As indicated earlier, the speed of the in-
withCD.Again,findingswereconsistentacrossstudies, hibitoryprocessisoneoftheparametersthatdetermines
bothintermsofeffectsizesandsignificancelevels. the IF. In other words, poor capability for response
Meta-analysis revealed no differences between IF- inhibition may reflect a relatively slow SSRT (Logan,
slopes of children with comorbid AD(cid:125)HD(cid:173)CD and 1994). The meta-analytic results for SSRT appear in
childrenwithAD(cid:125)HDalone.However,theseresultsneed Table4.Acrosssevenstudies,AD(cid:125)HDchildrenwereon
tobeinterpretedwithcautiongiventheinconsistencyin average103msecslowerthancontrolchildren.Amedium
botheffectsizesandsignificancelevels.Infact,twostud- combined effect size of d(cid:175).64 was obtained, which
iesdemonstratednosignificantdifferencesbetweenboth translates into 40% nonoverlap between the two group
groups (Daugherty et al., 1993; Pliszka & Borcherding, distributions.Botheffectsizesandsignificancelevelswere
1995), whereas one study indicated that children with homogeneousacrossstudies.
AD(cid:125)HD alone had flatter IF-slopes than comorbid Similarresultswerenotedforthecomparisonbetween
AD(cid:125)HD(cid:173)CDchildren(Schachar&Logan,1990). CDandcontrolchildren.Acrossfourstudies,SSRTwas
In three studies children with anxiety disorders were onaverageabout18msecslowerinCDchildrenthanin
compared with controls. Contrary to the predictions, controls.Theaverageeffectsizeofd(cid:175).51equalsCohen’s
childrenwithanxietydisordersdidnotexhibitsteeperIF- threshold for a medium effect. This effect size indicates
slopes than controls. Note also that these meta-analytic 34%nonoverlap between the SSRTdistributions ofthe
resultswereconsistentacrossstudies. twogroups.Diffusetestsindicatedheterogeneityinterms
RESPONSEINHIBITIONINCHILDPSYCHOPATHOLOGY 419
Table5
InhibitionFunctionSlopePlottedasaFunctionofZRFT(ZRFT–slope)
Samplesize Weightedmean
Groupcomparison Group1 Group2 Studies Group1 Group2 d CombinedZ
AD(cid:125)HD–Controls 99 117 6 21.0 23.8 0.19 1.31
CD–Controls 35 42 3 19.3 20.7 0.04 0.20
AD(cid:125)HD–CD 35 37 3 19.3 17.7 0.07 0.04
AD(cid:125)HD(cid:173)CD–AD(cid:125)HD 33 48 3 22.1 20.6 0.12 0.52
Anxious–Controls 42 42 3 21.1 20.7 0.03 0.26
ZRFT(cid:175)azscorewhichrepresentstherelativefinishingtimeoftheinhibitoryprocessandthe
responseexecutionprocessinSDunits(adiscussionofthismeasureisprovidedintheAppendix).
Diffuse comparisons of effect sizes and significance levels indicated homogeneous results for all
groupcomparisons.AllZscoresnonsignificant.
ofeffectsizesandsignificancelevels.Infact,threestudies impairments in response inhibition were not unique to
supported the direction of the meta-analytic result, and children with AD(cid:125)HD. Moreover, AD(cid:125)HD and CD
one study revealed faster SSRTs in children with CD childrendidnotdifferinthedegreeoftheirimpairments.
(Schachar&Tannock,1995).However,onlyOosterlaan No differences were noted between these two groups in
and Sergeant (1996) found that CD children had sig- terms of the inhibition function, nor in terms of the
nificantlyslowerSSRTsthancontrols. inhibitoryprocess.Itshouldbenoted,however,thatthe
Although the speed of the inhibitory process was on findings for the speed of the inhibitory process differed
averagealmost100msecslowerinAD(cid:125)HDchildrenthan sharplybetweenstudies.
inCDchildren,thecombinedsignificancelevelandeffect Second,AD(cid:125)HDchildrenwithandwithoutCDneither
sizeindicatednodifferencesbetweenthetwogroups.This differedintheirabilitytoinhibitresponses,norinterms
finding stems from the sharply conflicting findings ob- of the inhibitory process. Thus, AD(cid:125)HD(cid:173)CD children
tained in each of the four studies. Two studies revealed didnotshowgreaterimpairmentsthanchildrenwithonly
substantiallyfasterSSRTsforCDchildren.Incontrast, AD(cid:125)HD. However, findings were inconsistent between
theothertwostudiesdemonstratedthatCDchildrenhad studiesfortheslopeoftheinhibitionfunction.
somewhat slower inhibitory processes. Given these in- Third, no evidence was found for the prediction of
consistencies, the meta-analytic findings need to be enhancedresponseinhibitioninanxiouschildren.Infact,
interpretedwithconsiderablecaution. anxiousand controlchildren couldnotbedifferentiated
ComorbidAD(cid:125)HD(cid:173)CDchildrendidnotdifferfrom onanyofthedependentmeasures.
childrenwithAD(cid:125)HDalone.Furthermore,meta-analy- Finally,thelatencyofrespondingontheprimarytask
sisrevealednodifferencesinSSRTbetweenanxiousand wasslowerforAD(cid:125)HDchildrenthanforcontrols.There
control children. Note also that these findings were were no differences in the speed of responding between
homogeneousbothintermsofeffectsizesandintermsof theothergroups.
significancelevels. Obviously our findings are limited by the restricted
ZRFT-slope. Table 5 presents the meta-analytic number of studies currently available. This holds in
results for ZRFT-slope. Combined effect sizes and particularfor the comparisonsbetween CD and control
significancelevelsindicatedthattherewerenodifferences children, between AD(cid:125)HD and CD children, and be-
between the contrasted groups. All groups showed tweenAD(cid:125)HD(cid:173)CDandAD(cid:125)HDchildren.Someofthe
comparable ZRFT-slopes, indicating that group diff- meta-analytic results for these group comparisons were
erences in the capability for response inhibition were heterogeneous. Clearly, these inconsistencies call for
neither related to the probability of triggering the moreresearch.Furthermore,weneedtomentionthatour
inhibitory process, nor to differences in variability in findingsforAD(cid:125)HDwerelimitedbythefactthatintwo
latencyoftheinhibitoryprocess(Logan,1994).Notealso studies some children with this disorder also showed
thateffectsizesandsignificancelevelswerehomogeneous. ODDorCDsymptomatology.Withtheselimitationsin
mind,wenowdiscusstheresultsofthemeta-analysis.
Summary of Results Discussion
Despiteimportantmethodologicaldifferencesbetween This paper presents the results of a meta-analysis of
studies, the present meta-analysis revealed fairly con- studies with the stop task. The aim was to investigate
sistent results. In this section, we briefly summarise our whether data gathered with the stop task supports the
findings. response inhibition deficit hypothesis for AD(cid:125)HD
First, the major finding was that, compared with (Barkley, 1994, 1997; Douglas, 1988, 1989; Newman &
controls, both AD(cid:125)HD and CD children demonstrated Wallace, 1993; Pennington & Ozonoff, 1996; Quay,
flatter inhibition functions, indicating poor response 1988a,b,1997;Wender,1972).Furthermore,weinvesti-
inhibition.Inbothgroups,poorresponseinhibitionwas gated whether impaired response inhibition is uniquely
relatedtoaslowinhibitoryprocess,althoughthefindings relatedtoAD(cid:125)HD(Barkley,1994,1997;Douglas,1988,
were inconsistent for CD children. In other words, 1989; Pennington & Ozonoff, 1996; Wender, 1972) or
420 J.OOSTERLAANetal.
whetherdeficitsinresponseinhibitionarealsoevidentin Asimilarexplanationforresponseinhibitiondeficitsin
other psychopathological disorders. Finally, we ex- CDchildrenseemslesslikely.Althoughthereissupport
amined the suggestion that anxiety disorders are asso- forthepresenceofdeficitsinexecutivefunctioninginCD,
ciatedwithabnormallyhighlevelsofresponseinhibition Pennington and Ozonoff (1996) pointed out that these
(Quay,1988a,b). findings may be attributed to the presence of comorbid
Consistent and robust evidence was found for a AD(cid:125)HD.Recentstudiesthatcontrolledforthepresence
response inhibition deficit in AD(cid:125)HD. Relative to con- ofAD(cid:125)HDyieldconflictingfindings.Somestudieshave
trols, AD(cid:125)HD children demonstrated flatter inhibition shown that CD is indeed associated with impaired
functions,indicatingpoorresponseinhibition.Impaired executive functions (Se(cid:33)guin, Pihl, Harden, Tremblay, &
response inhibition was related to a slow inhibitory Boulerice, 1995; White et al., 1994). Other studies,
process. Our findings support recent theoretical notions however, failed to do so (Linz, Hooper, Hynd, Isaac, &
of AD(cid:125)HD in which poor response inhibition is sug- Gibson, 1990). The present meta-analysis shows that
gestedtolieat the heartofthisdisorder (Barkley,1994, childrenwithCD,butwithoutAD(cid:125)HD,exhibitresponse
1997; Douglas, 1988, 1989; Newman & Wallace, 1993; inhibition deficits. Since response inhibition is regarded
Pennington & Ozonoff, 1996; Quay, 1988a,b, 1997; asoneofthekeyexecutivefunctions,ourfindingsaddto
Wender,1972). theresearchindicatingexecutivefunctionsdeficitsinCD.
However, particularly important is that the meta- Besidesanexplanationintermsofadeficitinexecutive
analysis did not support the notion that response in- functioning, poor response inhibition in AD(cid:125)HD chil-
hibition deficits are unique to AD(cid:125)HD (Barkley, 1994, drenmayalsobeexplainedinotherways,forexampleas
1997;Douglas,1988,1989;Pennington&Ozonoff,1996; reflectinganon-optimalactivationstate(see,forreview,
Wender,1972).SimilartoAD(cid:125)HDchildren,CDchildren Sergeant&VanderMeere, 1990,1991;VanderMeere,
had flatter inhibition functions and slower inhibitory 1996).Inthecognitiveenergeticalmodelof information
processes compared with controls. Furthermore, res- processing (Sanders, 1983, 1990), an optimal activation
ponseinhibitiondeficitsdidnotdistinguishchildrenwith is required to maintain adequate motor processing
AD(cid:125)HDfromchildrenwithCD,norfromchildrenwith (Sternberg, 1969a, b). That is, activation is concerned
comorbidAD(cid:125)HD(cid:173)CD;thethreegroupsshowedsimi- with the readiness to respond (Pribram & McGuiness,
lar deficits in terms of their inhibition functions and the 1975; Sanders, 1983, 1990). A non-optimal activation
latencyoftheinhibitoryprocess.Thesecollectivefindings state causes impaired motor processing, which results
suggest that response inhibition deficits characterise in difficulties for both the execution and inhibition of
childrenwithbehaviourlabelledasundercontrolledorex- responses. Thus, the non-optimal activation hypothesis
ternalising (Achenbach, 1991; Achenbach & Edelbrock, canaccountforourfindingthatAD(cid:125)HDchildrenhada
1978). This conclusion confirms the predictions derived slower response execution process, as well as for our
from Quay’s psychobiological model of child psycho- findingofaresponseinhibitiondeficitinthesechildren.
pathology(Quay,1988a,b,1993,1997). Extensive research supports the hypothesis that AD(cid:125)
The finding that both AD(cid:125)HD and CD children HD involves a non-optimal activation state (see, for
evidence response inhibition deficits is not surprising. review,Sergeant&VanderMeere,1990,1991;Vander
Thereis considerable overlap in the symptomatology of Meere,1996).Activationisinfluencedbythepresentation
these two disorders (American Psychiatric Association, rate of stimuli in a task (Sanders, 1983, 1990). A high
1994;Barkley,1990;Biedermanetal.,1991;Nottelman presentation rate of stimuli causes high levels of ac-
& Jensen, 1995; Russo & Beidel, 1994). In fact, it is tivation;asloweventratecauseslowlevelsofactivation.
controversial whether AD(cid:125)HD and CD are distinct A recent study by Van der Meere, Stemerdink, and
clustersofabnormalbehaviour(Fergusson,Horwood,& Gunning (1995) demonstrated that AD(cid:125)HD children
Lloyd, 1991; Hinshaw, 1987; Hinshaw, Lahey, & Hart, were less able to inhibit inappropriate responding than
1993). Furthermore, AD(cid:125)HD has been found to be controlswithfastandslowpresentationratesofstimuli.
strongly predictive for the later development of CD However, AD(cid:125)HD children did not differ from their
(McGee, Williams, & Feehan, 1992; Taylor, Chadwick, controlpeersinamediumstimuluspresentationrate.
Heptinstall,&Dankaerts,1996).Consequently,itseems Particularlycompellingforthenon-optimalactivation
plausiblethatthetwodisorderssharecommondeficits,as state hypothesis is the research with methylphenidate.
wasfoundinthepresentresearch. Stimulants have been found to influence the activation
Although AD(cid:125)HD and CD share the impairment in level (Sanders, 1983, 1990; Sergeant & Van der Meere,
responseinhibition,thisimpairmentdoesnotnecessarily 1991).Interestingly,researchbyTannockandassociates
reflectthesamedysfunction.Poorresponseinhibitionin with the stop task has shown that methylphenidate
AD(cid:125)HDchildrenmaybepartofageneralimpairmentin ameliorates response inhibition deficits in AD(cid:125)HD chil-
executivefunctions,whichinturnmaybeattributableto dren(Tannocketal.,1989;Tannock,Schachar,&Logan,
a frontal lobe dysfunction (Lezak, 1983; Pennington & 1995). Clearly, these findings warrant more research to
Ozonoff, 1996; Torgesen, 1994). Recently, such an investigate the relationship between response inhibition
accountofAD(cid:125)HDhasbeenadvancedbyBarkley(1994, andactivationstate.
1997) and by Pennington and Ozonoff (1996). Indeed, PoorresponseinhibitioninAD(cid:125)HDchildrenmayalso
AD(cid:125)HD children have been found to demonstrate be understood as a motivational deficit. That is to say,
impairments on a variety of tests purported to measure these children do not expend the effort necessary to
frontal lobe functioning (see for review, Barkley et al., achieve and maintain optimal performance. In support
1992;Pennington&Ozonoff,1996;seealsoGrodzinsky of this view, it has been found that the performance of
&Diamond,1992;Shue&Douglas,1992). AD(cid:125)HDchildrenreliesmorestronglyonthepresenceof