- Email: [email protected]
Neuropsychological Executive Functions and DSM-IV ADHD Subtypes
Neuropsychological Executive Functions and DSM-IV ADHD Subtypes JOEL T. NIGG, PH.D., LISA G. BLASKEY, M.A., CYNTHIA L. HUANG-POLLOCK, M.A., AND MARSHA D. RAPPLEY, M.D.
ABSTRACT Objective: To evaluate and compare a focused set of component neuropsychological executive functions in the DSMIV attention-deficit/hyperactivity disorder combined (ADHD-C) and inattentive (ADHD-I) subtypes. Method: The Stop task, Tower of London, Stroop task, Trailmaking Test, and output speed measures were completed by 105 boys and girls aged 7–12 classified as either DSM-IV ADHD-C (n = 46), ADHD-I (n = 18), or community control (n = 41). Results: Both subtypes had deficits on output speed. A group ⫻ gender interaction was observed on the Stop task: boys with ADHDC were impaired versus boys with ADHD-I, whereas girls in the two subtypes did not differ. The ADHD-C type had a deficit in planning. Neither ADHD group had a deficit in interference control per se, although they were slower than controls on the Stroop tasks. Conclusions: ADHD-I shares neuropsychological deficits with ADHD-C in the domain of output speed; on most domains the subtypes did not differ. Neuropsychological distinctions between these ADHD subtypes may be few, depending on which domain of executive functioning is assessed, and these distinctions differ by gender. In the case of boys, the two subtypes may be distinguished by the specificity of motor inhibition deficits to ADHD-C. J. Am. Acad. Child Adolesc. Psychiatry, 2002, 41(1):59–66. Key Words: attention-deficit/hyperactivity disorder, executive functions, inhibition, subtypes.
Neuropsychological theories of attention-deficit/hyperactivity disorder (ADHD) have an important role in describing functional deficits that underlie symptomatic behavior. These theories guide investigations of etiology by linking behavior, process or mechanism, and neural systems (Zametkin and Rapoport, 1987). They are central to current attempts to understand ADHD (Barkley, 1997) and may contribute to improved clinical evaluation of ADHD as process deficits are better described. Neuropsychological studies, particularly in the domain of executive functioning (Pennington, 1997), also can help in evaluating the validity of the ADHD subtypes. Hyperactivity/impulsivity and inattention were combined in a single symptom dimension in DSM-III-R. As Accepted August 7, 2001. From the Departments of Psychology (Nigg, Blaskey, Huang-Pollock) and Pediatrics and Human Development (Rappley), Michigan State University, East Lansing. This work was supported by NIMH grant R03-MH57244. The Lansing and East Lansing School Districts provided valuable assistance. Reprint requests to Dr. Nigg, Department of Psychology, Michigan State University, East Lansing, MI 48824-1117. 0890-8567/02/4101–0059䉷2001 by the American Academy of Child and Adolescent Psychiatry.
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
a result, although ADHD has been clearly associated with neuropsychological deficits (Barkley, 1997) that may extend to girls as well as boys (Seidman et al., 1997), neuropsychological research on ADHD often failed to distinguish the inattentive, hyperactive, and combined subtypes now described in DSM-IV. Yet earlier data from the DSM-III era raised the possibility that hyperactive and inattentive groups had different cognitive profiles (Cantwell and Baker, 1992). At the same time, a consistent pattern of results was elusive both across and within samples on the basis of the pre-DSM-IV criteria (Barkley and Grodzinsky, 1992; Doyle et al., 2000; Pennington and Ozonoff, 1996). These findings all suggest that further investigation of DSM-IV subtypes is needed. Thus, neuropsychological theories of ADHD focus on executive functions and behavioral inhibition in the combined subtype (Barkley, 1997; Quay, 1997), yet it remains unclear whether the DSM-IV inattentive subtype has distinct deficits (Barkley, 1997; Carlson and Mann, 2000). The few neuropsychological studies of DSM-IV subtypes have yielded mixed results. Faraone et al. (1998) found no cognitive differences between subtypes on IQ and academic measures. Houghton et al. 59
NIGG ET AL.
(1999) found no differences between inattentive and combined subtypes on the Wisconsin Card Sort Test (WCST), Matching Familiar Figures, Trailmaking, Stroop, and Tower of London, although the combined subtype had qualitatively larger deficits. However, in a large sample, Klorman et al. (1999) found that ADHD combined and hyperactive type children performed more poorly than ADHD inattentive type children on the Tower of Hanoi; results were equivocal on the WCST. Clarification of the neuropsychological similarities and differences of these two ADHD subtypes can contribute to understanding their etiological relationship and guide future theory development. Because executive function deficits are central to conceptions of ADHD combined type, careful consideration of components of the broad executive functions domain is essential (Barkley, 1997; Pennington and Ozonoff, 1996). Such a componential approach is also important because the term executive functions is preliminary and underspecified (Lyon and Krasnegor, 1996; Pennington and Ozonoff, 1996). Models of executive functions have been proposed from behavioral, cognitive, developmental, and neuropsychological perspectives (Lyon and Krasnegor, 1996). In the neuropsychological framework, executive functions refer to regulation of response to context and maintenance of behavior on goal (Pennington, 1997). Nearly all experts agree that the domain is nonunitary. Typically, performance measures are only modestly correlated (Pennington, 1997), and imaging data suggest specific neural activation patterns for subdomains such as detecting response conflict and suppressing responses (Cabeza and Nyberg, 1997). However, the precise specification of subdomains of executive functioning is not agreed on. The present study examined four putative components of executive functions, following in approximate terms an executive functions model outlined by Pennington (1997): motor inhibition, planning, interference control, and set-shifting. Output response speed was also assessed for its relevance to crucial associated functions, such as effort or arousal, that are also theorized to be impaired in ADHD (Sergeant et al., 1999). Analysis of these subdomains of executive function can help to provide a systematic understanding of the nature of the ADHD subtypes. Several imaging studies have suggested abnormalities of structure and activation in basal ganglia and prefrontal neural regions, relevant to executive control, in children and adults with ADHD (Castellanos, 1999), underscoring the importance of con60
tinued neuropsychological characterization of syndrome subtypes. In behavioral terms, a recent review and metaanalysis suggests that inhibitory control and planning, but not set-shifting, are impaired in the ADHD combined type (Pennington and Ozonoff, 1996). Stroop-type interference control, apart from output speed, has an unclear pattern of results that has led to conflicting conclusions about its status in ADHD (Nigg, 2001). Summary
The goal of the present study was to examine the DSMIV combined and inattentive subtypes on a focused, rational battery of selected, commonly used executive function measures. We considered that the inattentive and combined subtypes might reflect related but distinct syndromes. For the combined type, we expected to replicate prior findings that describe deficits in motor inhibition (assessed by a computerized Stop Signal task) and in planning (assessed by the Tower of London task). For the inattentive type, we predicted that deficits would emerge in set-shifting and interference control, consistent with greater problems in mental versus motor inhibition. For both subtypes, we predicted deficits in a vigilance/arousal system (Posner and Peterson, 1990) that would lead to slow motor output speed. METHOD Participants A total of 105 children participated in three groups: ADHD combined type (ADHD-C, n = 46), ADHD inattentive type (ADHD-I, n = 18), and non-ADHD controls (n = 41). Child participants from first through sixth grade were recruited from a local pediatric clinic specializing in ADHD referrals, from a local support group for parents of children with ADHD, and from invitation letters sent to parents of children in the local school districts. To the extent possible, control children were matched on sex, age, and recruitment source (a neighboring general pediatric clinic or the same local school districts). Diagnostic Assignment. Children were considered as possible ADHDC or ADHD-I if they either (1) passed prescreen cutoffs on both parent and teacher version of common ADHD rating instruments—Child Behavior Checklist or Teacher’s Report Form (Achenbach, 1991), Behavioral Assessment System for Children (Reynolds and Kamphaus, 1992), Conners (1997) Rating Scale, or a DSM-IV symptoms checklist (the Swanson et al. [1998] SNAP-IV or the ADHD Rating Scale); or (2) had been assigned a diagnosis of ADHD (any type) by a physician or psychologist in the community who used teacher and parent ratings to arrive at the diagnosis. Children were considered as possible controls if they were below cutoff scores on all of these parent and teacher scales and had never had ADHD diagnosed in the community. Diagnosis then was confirmed with parents by means of a structured diagnostic parent interview, the Diagnostic Interview Schedule
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ADHD SUBTYPES AND EXECUTIVE FUNCTIONS
for Children (DISC-IV) (Shaffer et al., 1997) supplemented by an “or” algorithm that followed the DSM-IV field trials validity data (Lahey et al., 1994). Prior versions of the DISC have exhibited acceptable reliability and validity. The “or” algorithm was as follows. If children met age of onset, duration, impairment, and cross-situational criteria, then diagnostic assignment was determined by summing parent-reported symptoms on the DISC-IV and teacher-reported symptoms on the DSM-IV symptom checklist. Supplemental teacher symptom endorsements in addition to the DISC-IV were used to reach diagnostic threshold for eight children (three ADHD-C, five ADHD-I; of these, one child required one added teacher symptom, and the others needed two teacher symptoms). Control children were negative for ADHD (all types) on the basis of the above criteria, with four or fewer symptoms in either domain as shown by the “or” algorithm. Cases with five symptoms of inattention or overactivity were excluded from all groups on the basis of the field trial data indicating that these borderline cases might have ADHD-C or ADHD-I (Lahey et al., 1994). Each child was assessed for reading disorder (RD). RD was assigned if (1) absolute level of average reading and spelling were less than or equal to a standard score of 85 and (2) Full Scale IQ minus reading/spelling average was at least 15 points. Full Scale IQ was estimated with a reliable and valid fivetest short form of the WISC-III (Wechsler, 1991): Vocabulary, Block Design, Information, Object Assembly, and Picture Completion (Sattler, 1992). Reading and spelling were assessed with the Wechsler Individual Achievement Test (Wechsler, 1992). Children were excluded from all groups if they had mental retardation, autistic disorder, Tourette’s disorder, current major depressive episode, bipolar disorder, or physical or neurological handicap ascertained by parent report. All children were native English speakers, had normal hearing and normal or corrected vision, and had a valid Full Scale IQ of more than 75. One child required estimation of IQ by the sample mean because he failed to complete IQ measures validly. Children who were prescribed stimulants were free of medication when tested. Prescribed stimulants included methylphenidate, dextroamphetamine, and the mixed-salts amphetamines. The average elapsed time since last dose was 22 hours for the short-acting preparations and 56 hours (minimum 48 hours) for the long-acting preparations. Eight children (n = 6 ADHD-C, n = 2 ADHD-I) were on other psychotropic medications, including bupropion (n = 5), pemoline (n = 1), carbamazepine (n = 1), and clonidine (n = 1). Covarying presence/absence of these medications had the same effect as removing these children from analyses, with essentially no change in results. Therefore, to maximize the representativeness of the sample, and because they met full criteria for ADHD, these children were retained in the sample. However, all results reflect covariation of medication status. Procedure After diagnostic assignment, children completed neuropsychological measures during a visit to Michigan State University. Test administrators were carefully trained, followed practiced scripts in providing test instructions, and were blind to child diagnostic assignment and study hypotheses. The following measures operationalized a componential executive model: Stop task (motor inhibition), Tower of London (planning), Stroop task (response conflict), and Trailmaking (set-shifting). Speed of motor output was indexed by Go reaction time (RT) on the Stop task, Stroop Color and Word naming, and Trailmaking A. We describe each measure in turn. Behavioral Inhibition. Behavioral inhibition requires the suppression of a prepotent motor response. It is postulated to require activation of a circuit linking basal ganglia and orbitoprefrontal cortex (Castellanos, 1999). We operationalized the process with the track-
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ing version of the Stop task by using the same procedures as Logan et al. (1997) and Nigg (1999). In brief, the Stop task is a computerized choice reaction-time task. After two blocks of 32 practice trials, four blocks of 64 trials were administered. In the tracking version of this task, Stop Signal RT, the index of inhibitory control, is estimated by subtracting mean Stop Signal latency from mean Go RT (Logan et al., 1997). Go RT served as an index of arousal or effort. For additional details of our procedure, see Nigg (1999). Planning. The term planning is used here to describe the ability to manipulate visual information in working memory. It was operationalized with the Tower of London procedure as described by Krikorian et al. (1994). Children were presented with boards containing three wooden pegs of unequal heights. Three wooden balls of different colors (red, blue, and green) were to be moved from a standard start position to match models displayed on an 8.5 ⫻ 11 inch paper. Rules included holding only one ball at a time and not taking any moves back. Children attempted 12 problems of increasing difficulty, beginning with two-move problems and ending with fivemove problems. Each problem was worth 3 points if solved on the first trial, 2 points if not solved until the second trial, 1 point if not solved until the third trial, and 0 points if not solved in three trials. This task and others like it require visualizing the solution several moves in advance and places demands on prefrontal cortex in adults and children (Levin et al., 1994). Prior studies suggest that similar tasks differentiate ADHD-C and control groups (Klorman et al., 1999). The outcome variable was the total score (range 0–36). Interference Control. Interference control refers to the ability to monitor response conflict and suppress a competing response in order to carry out a primary response. It entails activation of anterior cingulate and dorsolateral prefrontal cortex (Cabeza and Nyberg, 1997). It was operationalized with the Stroop Color-Word Interference Test (Golden, 1978). The Stroop is a widely used neuropsychological measure (MacLeod, 1991). The paper-and-pencil version of the task was administered, with 45 seconds per trial. To evaluate interference control rather than naming speed, we tested the interaction in a repeatedmeasures design of subtest (Color versus Color-Word) by group. Set-Shifting. The ability to rapidly alternate mental set was operationalized with the Trailmaking Test for Children, from the HalsteadReitan Neuropsychological Battery (Reitan, 1979). The Trailmaking Test requires children to trace a path between consecutive letters (Form A) and alternating letters and numbers (Form B) as rapidly as they can without making errors. The primary executive measure was time to complete Form B, whereas Form A time was viewed as an additional index of motor output speed. Response Speed. Output speed on timed tasks is often taken as an indication of the intactness of state moderators of performance, in particular effort (Sergeant et al., 1999) or vigilance/arousal (Posner and Peterson, 1990). To assess response speed, we used related tasks that required speed response but demanded less executive processing, namely the Go RT on the Stop task, Color Naming on the Stroop, and Trailmaking A. Those results are embedded along with results of each executive measure in their respective sections below. Data Reduction Missing Data and Outliers. Missing data occurred on 5.7% of the neuropsychological data points as a result of either examiner/computer error or child noncompliance. To ignore missing data is problematic (Cohen and Cohen, 1983), whereas substituting estimated scores assumes the data are missing completely at random (Rovine and Delaney, 1990). Therefore, following Cohen and Cohen (1983), we created a dummy-coded missing data variable (missing data on 0, 1, 2, or 3 variables) and estimated missing scores with a regression
61
NIGG ET AL.
model using age, IQ, sex, and group. The missing data variable was then covaried in all models; this ensured that results were independent of effects of imputed data (we do not discuss this variable in Results because of its always negligible effect). Following current methodological principals (Wilcox et al., 1998), extreme outliers (z > 4.0 in either direction) were truncated to either 0.5 SD beyond the next most extreme score (if that score was z < 3.5), or to z = 4.0, then z = 4.1 (and so on), preserving their rank order (1% of scores were adjusted: Stop Signal RT, n = 2, Tower of London, n = 1, Trailmaking A, B, n = 4). Validity of Componential Executive Function Model. Some conceptual (and thus empirical) overlap among executive measures is expected (e.g., working memory demand on the Stop task, inhibitory demand on the Tower task). However, the measures are expected to tap sufficiently distinct domains that their separate evaluation is warranted. Consistent with this model, the four primary executive measures had correlations ranging in absolute value from r = 0.17 (Tower of London with Trailmaking B) to r = 0.33 (Tower of London with Stop Signal RT). This indicated that no two primary measures shared more than about 10% of their variance and justified treating them as separate executive domains, even though there was some modest intercorrelation present. Correlations ranged higher for the response speed measures across tasks, with the highest being r = 0.41 (Go RT with Stroop Word Naming). As might be expected by their similar demands, within-task correlations reached r = 0.47 for Trailmaking A and B, and r = .61 for Stroop Color and Word Naming. Overall, the intercorrelations were consistent with the executive function literature (Pennington, 1997). Plan of Data Analysis. After an omnibus multivariate analysis of covariance (MANCOVA) (with age, missing data, medication status) to ascertain that group effects were present in the data, the initial analysis for each variable consisted of a 2 ⫻ 2 ANCOVA of group (control, ADHD, ADHD-I) ⫻ gender of each primary executive task
to locate the group effects (Keppel, 1991). Checking possible gender effects is important because of uncertainty about similarity of deficits in boys and girls (Seidman et al., 1997). When the group ⫻ gender interaction was nonsignificant, boys and girls were combined so that group effects could be examined. Hypothesized planned comparisons were intended, with the ADHD-C group expected to have deficits on two tasks (Stop inhibition and Tower of London planning) and the ADHD-I group expected to show deficits on two other tasks (Stroop interference and Trailmaking B). Therefore, no correction was made for multiple comparisons (Keppel, 1991). However, as an extra precaution against type-I error, two-group comparisons were only conducted if the omnibus three-group comparison was significant (Keppel, 1991). Power and Effect Size. We report eta square (η2, similar to r2; Cohen and Cohen, 1983) as an index of effect size. When η2 = 0.15, effects are “large” in magnitude, and when η2 = 0.06, effects are “medium” in magnitude (Cohen and Cohen, 1983). Assuming the usual mediumto-large effect sizes in the ADHD literature of approximately η2 = 0.12 or d = 0.65 (Pennington and Ozonoff, 1996), the present study had power of 0.93 for the ADHD-C versus control comparisons, 0.81 for the ADHD-I versus control comparisons, and 0.84 for the ADHDC versus ADHD-I comparisons (all two-tailed). Power for the threegroup omnibus was 0.80 to detect η2 = 0.09. RESULTS Sample Description
Sample descriptive characteristics are summarized in Table 1. As shown, groups did not differ in age or gender composition.
TABLE 1 Demographics Dependent Measure
Comparison
n Age (months) % Male Ethnicity (% white) % Reading disorder % ODD % Conduct disorder Full Scale IQ Reading standard score DSM-IV hyperactive symptoms DSM-IV inattentive symptoms BASC Attention Problems CBCL Attention Problems BASC Conduct Problems CBCL Delinquency BASC Aggression CBCL Aggression CBCl Anxiety-Depression
41 121.4 (12.0) 59 68 0 2 0 109.4 (12.9) 108.7 (15.2) 0.37 (0.9) 0.89 (1.5) 44.8 (7.5) 52.5 (5.0) 42.8 (5.3) 53.8 (5.2) 43.1 (8.0) 52.5 (5.1) 54.6 (7.7)
ADHD-C 46 115.1 76 91 7 65 22 101.5 97.7 7.8 8.0 69.7 75.0 71.5 66.3 69.3 71.9 68.0
(18.1)
(11.6) (13.6) (1.2) (1.2) (8.7) (10.5) (16.6) (9.5) (15.1) (12.5) (14.4)
ADHD-I 18 123.7 67 83 17 22 0 104.9 98.1 1.9 7.3 67.0 59.4 51.4 54.5 58.8 54.5 57.8
(12.0)
(17.8) (13.5) (1.1) (1.8) (3.7) (9.8) (7.5) (6.3) (10.3) (5.5) (7.1)
Significance ( p) NS NS <.03a <.04b <.001a,b,c .001a,c .03a <.002a,b <.001a,b,c <.001a,b <.001a,b <.001a,b,c <.001a,b,c <.001a,c <.001a,b <.001a,c .001a,c
Note: ADHD-C = attention-deficit/hyperactivity disorder combined type; ADHD-I = ADHD inattentive type; ODD = oppositional defiant disorder; BASC = Behavioral Assessment System for Children; CBCL = Child Behavior Checklist; NS = not significant. DSM-IV symptoms represent parent Diagnostic Interview Schedule for Children plus teacher endorsement (“or” algorithm). Behavior ratings represent maternal T scores. Superscript letters indicate significant pairwise differences: a = ADHD-C vs. control, b = ADHD-I vs. control, and c = ADHD-C vs. ADHD-I. Standard deviations in parentheses.
62
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ADHD SUBTYPES AND EXECUTIVE FUNCTIONS
There was a tendency for more ethnic minority representation in the control group. The ADHD-C and ADHD-I groups did not differ on IQ, but the ADHDC group had a lower IQ than the control group. As expected, the ADHD-C group had a greater ratio of conduct and oppositional defiant disorders than the other two groups, RD was more common in the ADHD-I than in the control group, and academic achievement was weaker in the two ADHD groups than in the controls. The behavioral ratings reflect the expected patterns of disturbance based on the diagnostic subtypes. Controversy exists regarding whether IQ should be covaried in cognitive studies of ADHD. For simplicity, we present results without IQ covaried. However, all analyses were checked after covarying IQ, RD, and oppositional and conduct disorder symptoms, with no change in results unless noted. Omnibus MANCOVA
The omnibus MANCOVA of the primary neuropsychological variables revealed a large group effect, with Wilks λ = 0.67, and F14,186 = 2.93, p < .001, η2 = 0.18. We therefore examined each executive domain in turn, beginning with a 2 ⫻ 2 ANCOVA of group ⫻ gender. Raw scores for the neuropsychological measures are shown in Table 2. Behavioral Inhibition: Stop Signal Reaction Time
Stop Signal RT yielded a significant group ⫻ gender interaction (F2,96 = 4.75, p = .011, η2 = 0.09). We therefore examined group ⫻ gender interactions for each pairwise comparison (Keppel, 1991). When the ADHD-C group was compared with the control group, there was no interaction (p = .62), so boys and girls were combined. Stop Signal RTs were markedly slower in the ADHD-C
group than in the control group (F1,81 = 21.4, p < .001, η2 = 0.21); the effect was robust to covariates. When the ADHD-I group was compared with the control group, the group ⫻ gender interaction was significant (F1,51 = 10.99, p = .013, η2 = 0.12), so boys and girls were examined separately (see Table 2 footnote for gender means). Those analyses showed no deficit for the ADHD-I boys (F < 1.0, p = .74, η2 < 0.01), but a significant deficit for the ADHD-I girls (F1,17 = 13.3, p = .002, η2 = 0.42). Finally, the ADHD-C versus ADHD-I analysis also revealed a group ⫻ gender interaction (F1,57 = 16.9, p = .007, η2 = 0.12), so boys and girls were again examined separately. ADHD-C boys were slower than ADHD-I boys (F1,42 = 4.29, p = .044, η2 = 0.09) (with IQ covaried, p = .05), whereas girls with ADHD-I and girls with ADHD-C did not differ (p = .15). Go RT revealed no group ⫻ gender interaction, (F < 1.0, p = .84). The group effect was significant (F2,99 = 4.78, p = .01, η2 = 0.09). As predicted, both ADHD subtypes had slow Go RT. The ADHD-C group had a modest but significant deficit (F1,81 = 4.39, p = .039, η2 = 0.052). The ADHD-I group had a slightly larger response speed deficit (F1,53 = 5.46, p = .023, η2 = 0.093). However, the two ADHD groups did not differ significantly (p = .18, η2 = 0.03; Table 2). Planning: Tower of London Total Points
The group ⫻ gender interaction was nonsignificant (p = .91). When boys and girls were combined, the effect of diagnostic group was significant (F2,99 = 6.49, p = .002, η2 = 0.12). Compared with the control group, the ADHD-C group had a large deficit on the Tower of London (F1,82 = 15.7, p < .001, η2 = 0.16). The ADHD-I group had a
TABLE 2 Means (SDs) on Neuropsychological Measures for Three Groups Group Dependent Measure Stop Task Stop Signal RT Stop Task Go RT Tower of London points Stroop Word Stroop Color Stroop Color-Word Trailmaking A time Trailmaking B time
Control
ADHD-C
ADHD-I
297.7 (82.8) 654.2 (129.1) 28.7 (2.7) 74.2 (12.1) 50.3 (8.9) 29.8 (9.3) 18.3 (6.9) 41.3 (19.0)
415.3 (140.7) 705.6 (104.4) 26.6 (2.8) 60.4 (12.1) 43.0 (8.7) 24.2 (7.8) 22.0 (10.3) 56.6 (36.8)
382.2 (200.1) 732.5 (107.2) 27.0 (3.1) 60.3 (13.2) 43.9 (11.0) 23.0 (5.9) 25.3 (12.1) 49.6 (21.4)
Note: ADHD-C = attention-deficit/hyperactivity disorder combined type; ADHD-I = ADHD inattentive type; RT = reaction time. For Stop Signal inhibition, means (SDs) for boys and girls were as follows. Controls: boys = 303.3 (89), girls = 292.0 (72); ADHD-C: boys = 426.4 (138), girls = 396.7 (126); ADHD-I: boys = 323.9 (177), girls = 471.3 (144).
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
63
NIGG ET AL.
smaller deficit that was shy of significance (F1,54 = 3.55, p = .08, η2 = 0.06); it was reduced further with IQ covaried (p = .13). The ADHD subtypes did not differ (p = .69, η2 < 0.01). Interference Control: Stroop Test
For the Color-Word condition, the group ⫻ gender interaction was nonsignificant (p = .24). Collapsing across gender, the main effect of diagnostic group was significant (F2,99 = 5.84, p = .004, η2 = 0.11). Simple effects revealed impairment with a modest deficit on Color-Word Naming for ADHD-C (F1,81 = 4.96, p = .029, η2 = 0.06) and for ADHD-I (F1,53 = 9.04, p = .004, η2 = 0.15), but the two subtypes did not differ significantly (p = .25, η2 = 0.02). Similar effects held for Word and Color Naming (Table 2). However, the most important question was whether we would find a specific interference effect, which we evaluated with a 2 ⫻ 2 repeated-measures ANCOVA of Color-Word and Color Naming speeds. All interaction terms were well short of significance (Fs < 1.0), with effect sizes in the trivial range (η2 < 0.01), indicating that neither ADHD subtype had a specific deficit in interference control apart from their shared deficit in slow naming speed. Results were the same for the interference score (Golden, 1978) and the residual Color-Word score regressed on Color Naming. Set-Shifting: Trailmaking
Group ⫻ gender interactions were nonsignificant for both Trailmaking A and Trailmaking B (both ps > .45). Collapsing across gender, Trailmaking B time failed to yield a significant main effect for diagnosis (p = .21, η2 = 0.03; Table 2). However, the simple output speed test, Trailmaking A, yielded a significant diagnostic group effect (F2,99 = 3.71, p = .028, η2 = 0.07). On that task, the ADHD-C group, as expected, showed no significant deficit (p = .33, η2 = 0.012), consistent with the prior literature (Pennington and Ozonoff, 1996). The ADHD-I group, in contrast, showed a sizeable deficit (F1,54 = 7.76, p = .007, η2 = 0.13) that remained significant with IQ covaried (p = .01) but fell shy of significance (p = .06) with RD covaried. The ADHD-C versus ADHD-I comparison was initially shy of significance (p = .06, η2 = 0.06) but was significant with IQ covaried (F1,58 = 4.31, p = .042, η2 = 0.07). DISCUSSION
This study investigated two ADHD subtypes on four operational domains of executive functioning and on out64
put speed. Overall, the findings are consistent with the literature, but they also introduce new considerations for clarifying the relation of the two ADHD subtypes. The two ADHD groups had similar deficits compared with controls on key response speed measures, such as speed of the Go response (Stop task) and naming speed on the Stroop subtests, suggesting shared deficits in effort or vigilance (Sergeant et al., 1999). Only the ADHD-C group had a deficit in planning, but even on that measure, the two ADHD subtypes did not differ significantly from one another. The finding that, on most comparisons, ADHD subtypes did not differ significantly was also reported by Houghton et al. (1999), as well as by Hinshaw et al. (under review) in a study of girls. The findings suggest that these two subtypes are related by virtue of some shared neuropsychological deficits. However, this conclusion depends on consideration of executive function component functions and gender. First, the two ADHD subtypes did differ on the Stop task. Our findings break new ground in comparing the two subtypes on motor inhibition. This Stop task is of substantial theoretical importance because of its isolation of a motor inhibition process, which is crucial to theories of ADHD (Barkley, 1997) but is not well-assessed in most executive tasks. Effects, however, depended on gender. For the ADHD-C group, both boys and girls exhibited deficits compared with controls. In addition, ADHD-C boys were impaired compared with ADHDI boys, supporting a mechanism-based distinction between the subtypes for boys with regard to motor inhibition. For the ADHD-I group, boys performed nearly normally on the Stop task, whereas girls were impaired (versus controls; see means in Table 2 footnote). Second, the ADHD subtypes differed on Trailmaking A, in which the ADHD-I group was impaired compared with the ADHD-C group, with IQ covaried. This finding is notable mainly because it supports the conclusion from the other measures that, like ADHD-C, ADHD-I is also characterized by slow motor output speed or impaired vigilance processes. One study failed to find a deficit in ADHDI on Trailmaking A (Houghton et al., 1999); the effect may depend on group selection procedures. Our failure to find deficits in ADHD-C on Trailmaking B echoes the prior literature (Pennington and Ozonoff, 1996). The finding of an ADHD-C (but not ADHD-I) deficit on planning replicates the findings of Klorman et al. (1999), although the direct subtype comparison was not significant in our smaller sample. We did not find a deficit in J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ADHD SUBTYPES AND EXECUTIVE FUNCTIONS
either ADHD group in the interference function of the Stroop task apart from naming speed. Many Stroop studies of ADHD failed to report the task ⫻ group interaction or interference score (e.g., Houghton et al., 1999). Like us, those that did so failed to find a significant ADHD deficit in children (e.g., Grodzinsky and Diamond, 1992). However, an interference control deficit has been reported in adolescents (MacLeod and Prior, 1996). Although our results were unchanged with reading ability covaried, more examination of the possible role of subclinical reading difficulties in concealing an ADHD interference control deficit in childhood will be of interest. By adolescence, even slow readers may show the interference effect in ADHD.
appeared to be a parallel between rated symptoms of inattention and neuropsychological deficits in output speed. Conclusion
Results inform neuropsychological executive theories of ADHD subtype dysfunction. The data suggest that ADHD-C and ADHD-I are related disorders that share deficits in vigilance or effort functions and, on many measures, may differ only in severity. However, at least in boys, they may be distinguishable on the basis of response inhibition deficits found only in the ADHD-C group. REFERENCES
Limitations
Sample size was not equal for the two ADHD subtypes, so statistical tests may underestimate the deficits in the ADHD-I versus the ADHD-C group, or the small ADHD-I group could yield chance findings. In addition, because we did not intend to focus on gender differences, the number of girls in the study was small; clarification of gender effects will require a larger sample of boys and girls. Although the “or” algorithm maintains fidelity to the validity data in the DSM-IV field trials, it lowers the threshold for identifying both combined and inattentive subtypes versus other (albeit less well validated) methods of totaling the symptoms. Children on long-acting medications were allowed into the sample (with no change in results); this may be a difference from some other ADHD samples. Also, it should be noted that the executive constructs were generally assessed with only one measure each; convergence across measures will be important to examine in future work. Clinical Implications
The sample was recruited in the community. Although inferential advantages support such sampling, and most ADHD children had received clinical services, confirmation that findings extend to clinic-recruited samples is needed. From a clinical perspective, the possibility that children with ADHD-I have deficits in response speed (indexing effort or arousal effects) that are similar to those of children with ADHD-C underscores the importance of recognizing that children with ADHD-I might have notable cognitive deficits that should be assessed. Whereas the ADHD-I group appears as a milder behavioral variant compared with the ADHD-C group, as a result of less severe disruptive symptoms, in both groups there J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
Achenbach TM (1991), Manual for the Child Behavior Checklist/418 and 1991 Profile. Burlington: University of Vermont Department of Psychiatry Barkley RA (1997), Behavioral inhibition, sustained attention, and executive function: constructing a unified theory of ADHD. Psychol Bull 121:65–94 Barkley RA, Grodzinsky GM (1992), Are tests of frontal lobe functions useful in the diagnosis of attention deficit disorders? Clin Neuropsychol 8:121–139 Cabeza R, Nyberg L (1997), Imaging cognition: an empirical review of PET studies with normal subjects. J Cogn Neurosci 9:1–26 Cantwell DP, Baker L (1992), Attention deficit disorder with and without hyperactivity: a review and comparison of matched groups. J Am Acad Child Adolesc Psychiatry 31:432–438 Carlson CL, Mann M (2000), Attention deficit hyperactivity disorder, predominantly inattentive subtype. Child Adolesc Psychiatr Clin N Am 9:499–510 Castellanos FX (1999), The psychobiology of attention-deficit/hyperactivity disorder. In: Handbook of Disruptive Behavior Disorders, Quay HC, Hogan AE, eds. New York: Kluwer/ Plenum, pp 179–198 Cohen J, Cohen P (1983), Applied Multivariate Regression/Correlation Analysis for the Behavioral Sciences, 2nd ed. Hillsdale, NJ: Erlbaum Conners CK (1997), Conners Rating Scales-Revised. Toronto: Multi-Health Systems Doyle AE, Biederman J, Seidman LJ, Weber W, Faraone SV (2000), Diagnostic efficiency of neuropsychological test scores for discriminating boys with and without attention deficit-hyperactivity disorder. J Consult Clin Psychol 68:477–488 Faraone SV, Biederman J, Weber W, Russell RL (1998), Psychiatric, neuropsychological and psychosocial features of DSM-IV subtypes of attention-deficit/hyperactivity disorder: results from a clinically referred sample. J Am Acad Child Adolesc Psychiatry 37:185–193 Golden CJ (1978), Stroop Color and Word Test: A Manual for Clinical and Experimental Uses. Chicago: Stoelting Grodzinsky GM, Diamond R (1992), Frontal lobe functioning in boys with attention deficit hyperactivity disorder. Dev Neuropsychol 8:427–445 Houghton S, Douglas G, West J et al. (1999), Differential patterns of executive function in children with attention-deficit hyperactivity disorder according to gender and subtype. J Child Neurol 14:801–805 Keppel G (1991), Design and Analysis, 3rd ed. Englewood Cliffs, NJ: Prentice Hall Klorman R, Hazel-Fernandez LA, Shaywitz SE et al. (1999), Executive functioning deficits in attention-deficit hyperactivity disorder are independent of oppositional defiant disorder or reading disorder. J Am Acad Child Adolesc Psychiatry 38:1148–1155 Krikorian R, Bartok J, Gay N (1994), Tower of London procedure: a standard method and developmental data. J Clin Exp Neuropsychol 16:840–850 Lahey BB, Applegate B, McBurnett K et al. (1994), DSM-IV field trials for attention-deficit/hyperactivity disorder in children and adolescents. Am J Psychiatry 151:1637–1685
65
NIGG ET AL.
Levin HS, Mendelsohn D, Lilly MA et al. (1994), Tower of London performance in relation to magnetic resonance imaging following closed head injury in children. Neuropsychology 8:171–179 Logan GD, Schachar RJ, Tannock R (1997), Impulsivity and inhibitory control. Psychol Sci 8:60–64 Lyon GR, Krasnegor NA (1996), Attention, Memory and Executive Function. Baltimore: Paul H. Brooks Publishing Company MacLeod CM (1991), Half a century of research on the Stroop effect: an integrative review. Psychol Bull 109:163–203 MacLeod D, Prior M (1996), Attention deficits in adolescents with ADHD and other clinical groups. Child Neuropsychol 2:1–10 Nigg JT (1999), The ADHD response inhibition deficit as measured by the Stop Task: replication with DSM-IV combined type, extension, and qualification. J Abnorm Child Psychol 27:391–400 Nigg JT (2001), Is ADHD an inhibitory disorder? Psychol Bull 127:571–598 Pennington BF (1997), Dimensions of executive functions in normal and abnormal development. In: Development of the Prefrontal Cortex: Evolution, Neurobiology, and Behavior, Krasnegor NA, Lyon GR, Goldman-Rakic P, eds. Baltimore: Paul H Brooks Publishing Company, pp 265–281 Pennington BF, Ozonoff S (1996), Executive functions and developmental psychopathology. J Child Psychol Psychiatry 37:51–87 Posner MI, Peterson SE (1990), The attention system of the human brain. Annu Rev Neurosci 13:25–42 Quay HC (1997), Inhibition and attention deficit hyperactivity disorder. J Abnorm Child Psychol 25:7–13 Reitan RM (1979), Manual for Administration of Neuropsychological Test Batteries for Adults and Children. Tucson, AZ: Author Reynolds CR, Kamphaus RW (1992), Behavioral Assessment System for Children Manual. Circle Pines, MN: American Guidance Service
66
Rovine MJ, Delaney M (1990), Missing data estimation in developmental research. In Statistical Methods in Longitudinal Research, Vol 1, von Eye A, ed. San Diego: Academic Press, pp 35–79 Sattler JM (1992), Assessment of Children: Revised and Updated 3rd Edition. San Diego: Author Seidman LJ, Biederman J, Faraone SV, Weber W (1997), A pilot study of neuropsychological function in girls with ADHD. J Am Acad Child Adolesc Psychiatry 36:366–373 Sergeant JA, Oosterlaan J, van der Meere J (1999), Information processing and energetic factors in attention deficit/hyperactivity disorder. In: Handbook of Disruptive Behavior Disorders, Quay HC, Hogan A, eds. New York: Kluwer/Plenum, pp 75–104 Shaffer D, Fisher P, Lucas C (1997), NIMH Diagnostic Interview Schedule for Children-IV. New York: Ruane Center for Early Diagnosis, Division of Child Psychiatry, Columbia University Swanson JM, Lerner MA, March J, Gresham FM (1998), Assessment and intervention for attention-deficit/hyperactivity disorder in the schools: lessons from the MTA study. Pediatr Clin North Am 46:993–1009 Wechsler D (1991), Wechsler Intelligence Scale for Children, 3rd ed. San Antonio, TX: Psychological Corporation Wechsler D (1992), Wechsler Individual Achievement Test. San Antonio, TX: Psychological Corporation Wilcox RR, Keselman HJ, Kowalchuk RK (1998), Can tests for treatment group equality be improved? The bootstrap and trimmed means conjecture. Br J Math Stat Psychol 51:123–134 Zametkin AJ, Rapoport JL (1987), Neurobiology of attention-deficit disorder with hyper-activity: where have we come in 50 years? J Am Acad Child Adolesc Psychiatry 26:676–686
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ABSTRACT Objective: To evaluate and compare a focused set of component neuropsychological executive functions in the DSMIV attention-deficit/hyperactivity disorder combined (ADHD-C) and inattentive (ADHD-I) subtypes. Method: The Stop task, Tower of London, Stroop task, Trailmaking Test, and output speed measures were completed by 105 boys and girls aged 7–12 classified as either DSM-IV ADHD-C (n = 46), ADHD-I (n = 18), or community control (n = 41). Results: Both subtypes had deficits on output speed. A group ⫻ gender interaction was observed on the Stop task: boys with ADHDC were impaired versus boys with ADHD-I, whereas girls in the two subtypes did not differ. The ADHD-C type had a deficit in planning. Neither ADHD group had a deficit in interference control per se, although they were slower than controls on the Stroop tasks. Conclusions: ADHD-I shares neuropsychological deficits with ADHD-C in the domain of output speed; on most domains the subtypes did not differ. Neuropsychological distinctions between these ADHD subtypes may be few, depending on which domain of executive functioning is assessed, and these distinctions differ by gender. In the case of boys, the two subtypes may be distinguished by the specificity of motor inhibition deficits to ADHD-C. J. Am. Acad. Child Adolesc. Psychiatry, 2002, 41(1):59–66. Key Words: attention-deficit/hyperactivity disorder, executive functions, inhibition, subtypes.
Neuropsychological theories of attention-deficit/hyperactivity disorder (ADHD) have an important role in describing functional deficits that underlie symptomatic behavior. These theories guide investigations of etiology by linking behavior, process or mechanism, and neural systems (Zametkin and Rapoport, 1987). They are central to current attempts to understand ADHD (Barkley, 1997) and may contribute to improved clinical evaluation of ADHD as process deficits are better described. Neuropsychological studies, particularly in the domain of executive functioning (Pennington, 1997), also can help in evaluating the validity of the ADHD subtypes. Hyperactivity/impulsivity and inattention were combined in a single symptom dimension in DSM-III-R. As Accepted August 7, 2001. From the Departments of Psychology (Nigg, Blaskey, Huang-Pollock) and Pediatrics and Human Development (Rappley), Michigan State University, East Lansing. This work was supported by NIMH grant R03-MH57244. The Lansing and East Lansing School Districts provided valuable assistance. Reprint requests to Dr. Nigg, Department of Psychology, Michigan State University, East Lansing, MI 48824-1117. 0890-8567/02/4101–0059䉷2001 by the American Academy of Child and Adolescent Psychiatry.
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
a result, although ADHD has been clearly associated with neuropsychological deficits (Barkley, 1997) that may extend to girls as well as boys (Seidman et al., 1997), neuropsychological research on ADHD often failed to distinguish the inattentive, hyperactive, and combined subtypes now described in DSM-IV. Yet earlier data from the DSM-III era raised the possibility that hyperactive and inattentive groups had different cognitive profiles (Cantwell and Baker, 1992). At the same time, a consistent pattern of results was elusive both across and within samples on the basis of the pre-DSM-IV criteria (Barkley and Grodzinsky, 1992; Doyle et al., 2000; Pennington and Ozonoff, 1996). These findings all suggest that further investigation of DSM-IV subtypes is needed. Thus, neuropsychological theories of ADHD focus on executive functions and behavioral inhibition in the combined subtype (Barkley, 1997; Quay, 1997), yet it remains unclear whether the DSM-IV inattentive subtype has distinct deficits (Barkley, 1997; Carlson and Mann, 2000). The few neuropsychological studies of DSM-IV subtypes have yielded mixed results. Faraone et al. (1998) found no cognitive differences between subtypes on IQ and academic measures. Houghton et al. 59
NIGG ET AL.
(1999) found no differences between inattentive and combined subtypes on the Wisconsin Card Sort Test (WCST), Matching Familiar Figures, Trailmaking, Stroop, and Tower of London, although the combined subtype had qualitatively larger deficits. However, in a large sample, Klorman et al. (1999) found that ADHD combined and hyperactive type children performed more poorly than ADHD inattentive type children on the Tower of Hanoi; results were equivocal on the WCST. Clarification of the neuropsychological similarities and differences of these two ADHD subtypes can contribute to understanding their etiological relationship and guide future theory development. Because executive function deficits are central to conceptions of ADHD combined type, careful consideration of components of the broad executive functions domain is essential (Barkley, 1997; Pennington and Ozonoff, 1996). Such a componential approach is also important because the term executive functions is preliminary and underspecified (Lyon and Krasnegor, 1996; Pennington and Ozonoff, 1996). Models of executive functions have been proposed from behavioral, cognitive, developmental, and neuropsychological perspectives (Lyon and Krasnegor, 1996). In the neuropsychological framework, executive functions refer to regulation of response to context and maintenance of behavior on goal (Pennington, 1997). Nearly all experts agree that the domain is nonunitary. Typically, performance measures are only modestly correlated (Pennington, 1997), and imaging data suggest specific neural activation patterns for subdomains such as detecting response conflict and suppressing responses (Cabeza and Nyberg, 1997). However, the precise specification of subdomains of executive functioning is not agreed on. The present study examined four putative components of executive functions, following in approximate terms an executive functions model outlined by Pennington (1997): motor inhibition, planning, interference control, and set-shifting. Output response speed was also assessed for its relevance to crucial associated functions, such as effort or arousal, that are also theorized to be impaired in ADHD (Sergeant et al., 1999). Analysis of these subdomains of executive function can help to provide a systematic understanding of the nature of the ADHD subtypes. Several imaging studies have suggested abnormalities of structure and activation in basal ganglia and prefrontal neural regions, relevant to executive control, in children and adults with ADHD (Castellanos, 1999), underscoring the importance of con60
tinued neuropsychological characterization of syndrome subtypes. In behavioral terms, a recent review and metaanalysis suggests that inhibitory control and planning, but not set-shifting, are impaired in the ADHD combined type (Pennington and Ozonoff, 1996). Stroop-type interference control, apart from output speed, has an unclear pattern of results that has led to conflicting conclusions about its status in ADHD (Nigg, 2001). Summary
The goal of the present study was to examine the DSMIV combined and inattentive subtypes on a focused, rational battery of selected, commonly used executive function measures. We considered that the inattentive and combined subtypes might reflect related but distinct syndromes. For the combined type, we expected to replicate prior findings that describe deficits in motor inhibition (assessed by a computerized Stop Signal task) and in planning (assessed by the Tower of London task). For the inattentive type, we predicted that deficits would emerge in set-shifting and interference control, consistent with greater problems in mental versus motor inhibition. For both subtypes, we predicted deficits in a vigilance/arousal system (Posner and Peterson, 1990) that would lead to slow motor output speed. METHOD Participants A total of 105 children participated in three groups: ADHD combined type (ADHD-C, n = 46), ADHD inattentive type (ADHD-I, n = 18), and non-ADHD controls (n = 41). Child participants from first through sixth grade were recruited from a local pediatric clinic specializing in ADHD referrals, from a local support group for parents of children with ADHD, and from invitation letters sent to parents of children in the local school districts. To the extent possible, control children were matched on sex, age, and recruitment source (a neighboring general pediatric clinic or the same local school districts). Diagnostic Assignment. Children were considered as possible ADHDC or ADHD-I if they either (1) passed prescreen cutoffs on both parent and teacher version of common ADHD rating instruments—Child Behavior Checklist or Teacher’s Report Form (Achenbach, 1991), Behavioral Assessment System for Children (Reynolds and Kamphaus, 1992), Conners (1997) Rating Scale, or a DSM-IV symptoms checklist (the Swanson et al. [1998] SNAP-IV or the ADHD Rating Scale); or (2) had been assigned a diagnosis of ADHD (any type) by a physician or psychologist in the community who used teacher and parent ratings to arrive at the diagnosis. Children were considered as possible controls if they were below cutoff scores on all of these parent and teacher scales and had never had ADHD diagnosed in the community. Diagnosis then was confirmed with parents by means of a structured diagnostic parent interview, the Diagnostic Interview Schedule
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ADHD SUBTYPES AND EXECUTIVE FUNCTIONS
for Children (DISC-IV) (Shaffer et al., 1997) supplemented by an “or” algorithm that followed the DSM-IV field trials validity data (Lahey et al., 1994). Prior versions of the DISC have exhibited acceptable reliability and validity. The “or” algorithm was as follows. If children met age of onset, duration, impairment, and cross-situational criteria, then diagnostic assignment was determined by summing parent-reported symptoms on the DISC-IV and teacher-reported symptoms on the DSM-IV symptom checklist. Supplemental teacher symptom endorsements in addition to the DISC-IV were used to reach diagnostic threshold for eight children (three ADHD-C, five ADHD-I; of these, one child required one added teacher symptom, and the others needed two teacher symptoms). Control children were negative for ADHD (all types) on the basis of the above criteria, with four or fewer symptoms in either domain as shown by the “or” algorithm. Cases with five symptoms of inattention or overactivity were excluded from all groups on the basis of the field trial data indicating that these borderline cases might have ADHD-C or ADHD-I (Lahey et al., 1994). Each child was assessed for reading disorder (RD). RD was assigned if (1) absolute level of average reading and spelling were less than or equal to a standard score of 85 and (2) Full Scale IQ minus reading/spelling average was at least 15 points. Full Scale IQ was estimated with a reliable and valid fivetest short form of the WISC-III (Wechsler, 1991): Vocabulary, Block Design, Information, Object Assembly, and Picture Completion (Sattler, 1992). Reading and spelling were assessed with the Wechsler Individual Achievement Test (Wechsler, 1992). Children were excluded from all groups if they had mental retardation, autistic disorder, Tourette’s disorder, current major depressive episode, bipolar disorder, or physical or neurological handicap ascertained by parent report. All children were native English speakers, had normal hearing and normal or corrected vision, and had a valid Full Scale IQ of more than 75. One child required estimation of IQ by the sample mean because he failed to complete IQ measures validly. Children who were prescribed stimulants were free of medication when tested. Prescribed stimulants included methylphenidate, dextroamphetamine, and the mixed-salts amphetamines. The average elapsed time since last dose was 22 hours for the short-acting preparations and 56 hours (minimum 48 hours) for the long-acting preparations. Eight children (n = 6 ADHD-C, n = 2 ADHD-I) were on other psychotropic medications, including bupropion (n = 5), pemoline (n = 1), carbamazepine (n = 1), and clonidine (n = 1). Covarying presence/absence of these medications had the same effect as removing these children from analyses, with essentially no change in results. Therefore, to maximize the representativeness of the sample, and because they met full criteria for ADHD, these children were retained in the sample. However, all results reflect covariation of medication status. Procedure After diagnostic assignment, children completed neuropsychological measures during a visit to Michigan State University. Test administrators were carefully trained, followed practiced scripts in providing test instructions, and were blind to child diagnostic assignment and study hypotheses. The following measures operationalized a componential executive model: Stop task (motor inhibition), Tower of London (planning), Stroop task (response conflict), and Trailmaking (set-shifting). Speed of motor output was indexed by Go reaction time (RT) on the Stop task, Stroop Color and Word naming, and Trailmaking A. We describe each measure in turn. Behavioral Inhibition. Behavioral inhibition requires the suppression of a prepotent motor response. It is postulated to require activation of a circuit linking basal ganglia and orbitoprefrontal cortex (Castellanos, 1999). We operationalized the process with the track-
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ing version of the Stop task by using the same procedures as Logan et al. (1997) and Nigg (1999). In brief, the Stop task is a computerized choice reaction-time task. After two blocks of 32 practice trials, four blocks of 64 trials were administered. In the tracking version of this task, Stop Signal RT, the index of inhibitory control, is estimated by subtracting mean Stop Signal latency from mean Go RT (Logan et al., 1997). Go RT served as an index of arousal or effort. For additional details of our procedure, see Nigg (1999). Planning. The term planning is used here to describe the ability to manipulate visual information in working memory. It was operationalized with the Tower of London procedure as described by Krikorian et al. (1994). Children were presented with boards containing three wooden pegs of unequal heights. Three wooden balls of different colors (red, blue, and green) were to be moved from a standard start position to match models displayed on an 8.5 ⫻ 11 inch paper. Rules included holding only one ball at a time and not taking any moves back. Children attempted 12 problems of increasing difficulty, beginning with two-move problems and ending with fivemove problems. Each problem was worth 3 points if solved on the first trial, 2 points if not solved until the second trial, 1 point if not solved until the third trial, and 0 points if not solved in three trials. This task and others like it require visualizing the solution several moves in advance and places demands on prefrontal cortex in adults and children (Levin et al., 1994). Prior studies suggest that similar tasks differentiate ADHD-C and control groups (Klorman et al., 1999). The outcome variable was the total score (range 0–36). Interference Control. Interference control refers to the ability to monitor response conflict and suppress a competing response in order to carry out a primary response. It entails activation of anterior cingulate and dorsolateral prefrontal cortex (Cabeza and Nyberg, 1997). It was operationalized with the Stroop Color-Word Interference Test (Golden, 1978). The Stroop is a widely used neuropsychological measure (MacLeod, 1991). The paper-and-pencil version of the task was administered, with 45 seconds per trial. To evaluate interference control rather than naming speed, we tested the interaction in a repeatedmeasures design of subtest (Color versus Color-Word) by group. Set-Shifting. The ability to rapidly alternate mental set was operationalized with the Trailmaking Test for Children, from the HalsteadReitan Neuropsychological Battery (Reitan, 1979). The Trailmaking Test requires children to trace a path between consecutive letters (Form A) and alternating letters and numbers (Form B) as rapidly as they can without making errors. The primary executive measure was time to complete Form B, whereas Form A time was viewed as an additional index of motor output speed. Response Speed. Output speed on timed tasks is often taken as an indication of the intactness of state moderators of performance, in particular effort (Sergeant et al., 1999) or vigilance/arousal (Posner and Peterson, 1990). To assess response speed, we used related tasks that required speed response but demanded less executive processing, namely the Go RT on the Stop task, Color Naming on the Stroop, and Trailmaking A. Those results are embedded along with results of each executive measure in their respective sections below. Data Reduction Missing Data and Outliers. Missing data occurred on 5.7% of the neuropsychological data points as a result of either examiner/computer error or child noncompliance. To ignore missing data is problematic (Cohen and Cohen, 1983), whereas substituting estimated scores assumes the data are missing completely at random (Rovine and Delaney, 1990). Therefore, following Cohen and Cohen (1983), we created a dummy-coded missing data variable (missing data on 0, 1, 2, or 3 variables) and estimated missing scores with a regression
61
NIGG ET AL.
model using age, IQ, sex, and group. The missing data variable was then covaried in all models; this ensured that results were independent of effects of imputed data (we do not discuss this variable in Results because of its always negligible effect). Following current methodological principals (Wilcox et al., 1998), extreme outliers (z > 4.0 in either direction) were truncated to either 0.5 SD beyond the next most extreme score (if that score was z < 3.5), or to z = 4.0, then z = 4.1 (and so on), preserving their rank order (1% of scores were adjusted: Stop Signal RT, n = 2, Tower of London, n = 1, Trailmaking A, B, n = 4). Validity of Componential Executive Function Model. Some conceptual (and thus empirical) overlap among executive measures is expected (e.g., working memory demand on the Stop task, inhibitory demand on the Tower task). However, the measures are expected to tap sufficiently distinct domains that their separate evaluation is warranted. Consistent with this model, the four primary executive measures had correlations ranging in absolute value from r = 0.17 (Tower of London with Trailmaking B) to r = 0.33 (Tower of London with Stop Signal RT). This indicated that no two primary measures shared more than about 10% of their variance and justified treating them as separate executive domains, even though there was some modest intercorrelation present. Correlations ranged higher for the response speed measures across tasks, with the highest being r = 0.41 (Go RT with Stroop Word Naming). As might be expected by their similar demands, within-task correlations reached r = 0.47 for Trailmaking A and B, and r = .61 for Stroop Color and Word Naming. Overall, the intercorrelations were consistent with the executive function literature (Pennington, 1997). Plan of Data Analysis. After an omnibus multivariate analysis of covariance (MANCOVA) (with age, missing data, medication status) to ascertain that group effects were present in the data, the initial analysis for each variable consisted of a 2 ⫻ 2 ANCOVA of group (control, ADHD, ADHD-I) ⫻ gender of each primary executive task
to locate the group effects (Keppel, 1991). Checking possible gender effects is important because of uncertainty about similarity of deficits in boys and girls (Seidman et al., 1997). When the group ⫻ gender interaction was nonsignificant, boys and girls were combined so that group effects could be examined. Hypothesized planned comparisons were intended, with the ADHD-C group expected to have deficits on two tasks (Stop inhibition and Tower of London planning) and the ADHD-I group expected to show deficits on two other tasks (Stroop interference and Trailmaking B). Therefore, no correction was made for multiple comparisons (Keppel, 1991). However, as an extra precaution against type-I error, two-group comparisons were only conducted if the omnibus three-group comparison was significant (Keppel, 1991). Power and Effect Size. We report eta square (η2, similar to r2; Cohen and Cohen, 1983) as an index of effect size. When η2 = 0.15, effects are “large” in magnitude, and when η2 = 0.06, effects are “medium” in magnitude (Cohen and Cohen, 1983). Assuming the usual mediumto-large effect sizes in the ADHD literature of approximately η2 = 0.12 or d = 0.65 (Pennington and Ozonoff, 1996), the present study had power of 0.93 for the ADHD-C versus control comparisons, 0.81 for the ADHD-I versus control comparisons, and 0.84 for the ADHDC versus ADHD-I comparisons (all two-tailed). Power for the threegroup omnibus was 0.80 to detect η2 = 0.09. RESULTS Sample Description
Sample descriptive characteristics are summarized in Table 1. As shown, groups did not differ in age or gender composition.
TABLE 1 Demographics Dependent Measure
Comparison
n Age (months) % Male Ethnicity (% white) % Reading disorder % ODD % Conduct disorder Full Scale IQ Reading standard score DSM-IV hyperactive symptoms DSM-IV inattentive symptoms BASC Attention Problems CBCL Attention Problems BASC Conduct Problems CBCL Delinquency BASC Aggression CBCL Aggression CBCl Anxiety-Depression
41 121.4 (12.0) 59 68 0 2 0 109.4 (12.9) 108.7 (15.2) 0.37 (0.9) 0.89 (1.5) 44.8 (7.5) 52.5 (5.0) 42.8 (5.3) 53.8 (5.2) 43.1 (8.0) 52.5 (5.1) 54.6 (7.7)
ADHD-C 46 115.1 76 91 7 65 22 101.5 97.7 7.8 8.0 69.7 75.0 71.5 66.3 69.3 71.9 68.0
(18.1)
(11.6) (13.6) (1.2) (1.2) (8.7) (10.5) (16.6) (9.5) (15.1) (12.5) (14.4)
ADHD-I 18 123.7 67 83 17 22 0 104.9 98.1 1.9 7.3 67.0 59.4 51.4 54.5 58.8 54.5 57.8
(12.0)
(17.8) (13.5) (1.1) (1.8) (3.7) (9.8) (7.5) (6.3) (10.3) (5.5) (7.1)
Significance ( p) NS NS <.03a <.04b <.001a,b,c .001a,c .03a <.002a,b <.001a,b,c <.001a,b <.001a,b <.001a,b,c <.001a,b,c <.001a,c <.001a,b <.001a,c .001a,c
Note: ADHD-C = attention-deficit/hyperactivity disorder combined type; ADHD-I = ADHD inattentive type; ODD = oppositional defiant disorder; BASC = Behavioral Assessment System for Children; CBCL = Child Behavior Checklist; NS = not significant. DSM-IV symptoms represent parent Diagnostic Interview Schedule for Children plus teacher endorsement (“or” algorithm). Behavior ratings represent maternal T scores. Superscript letters indicate significant pairwise differences: a = ADHD-C vs. control, b = ADHD-I vs. control, and c = ADHD-C vs. ADHD-I. Standard deviations in parentheses.
62
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ADHD SUBTYPES AND EXECUTIVE FUNCTIONS
There was a tendency for more ethnic minority representation in the control group. The ADHD-C and ADHD-I groups did not differ on IQ, but the ADHDC group had a lower IQ than the control group. As expected, the ADHD-C group had a greater ratio of conduct and oppositional defiant disorders than the other two groups, RD was more common in the ADHD-I than in the control group, and academic achievement was weaker in the two ADHD groups than in the controls. The behavioral ratings reflect the expected patterns of disturbance based on the diagnostic subtypes. Controversy exists regarding whether IQ should be covaried in cognitive studies of ADHD. For simplicity, we present results without IQ covaried. However, all analyses were checked after covarying IQ, RD, and oppositional and conduct disorder symptoms, with no change in results unless noted. Omnibus MANCOVA
The omnibus MANCOVA of the primary neuropsychological variables revealed a large group effect, with Wilks λ = 0.67, and F14,186 = 2.93, p < .001, η2 = 0.18. We therefore examined each executive domain in turn, beginning with a 2 ⫻ 2 ANCOVA of group ⫻ gender. Raw scores for the neuropsychological measures are shown in Table 2. Behavioral Inhibition: Stop Signal Reaction Time
Stop Signal RT yielded a significant group ⫻ gender interaction (F2,96 = 4.75, p = .011, η2 = 0.09). We therefore examined group ⫻ gender interactions for each pairwise comparison (Keppel, 1991). When the ADHD-C group was compared with the control group, there was no interaction (p = .62), so boys and girls were combined. Stop Signal RTs were markedly slower in the ADHD-C
group than in the control group (F1,81 = 21.4, p < .001, η2 = 0.21); the effect was robust to covariates. When the ADHD-I group was compared with the control group, the group ⫻ gender interaction was significant (F1,51 = 10.99, p = .013, η2 = 0.12), so boys and girls were examined separately (see Table 2 footnote for gender means). Those analyses showed no deficit for the ADHD-I boys (F < 1.0, p = .74, η2 < 0.01), but a significant deficit for the ADHD-I girls (F1,17 = 13.3, p = .002, η2 = 0.42). Finally, the ADHD-C versus ADHD-I analysis also revealed a group ⫻ gender interaction (F1,57 = 16.9, p = .007, η2 = 0.12), so boys and girls were again examined separately. ADHD-C boys were slower than ADHD-I boys (F1,42 = 4.29, p = .044, η2 = 0.09) (with IQ covaried, p = .05), whereas girls with ADHD-I and girls with ADHD-C did not differ (p = .15). Go RT revealed no group ⫻ gender interaction, (F < 1.0, p = .84). The group effect was significant (F2,99 = 4.78, p = .01, η2 = 0.09). As predicted, both ADHD subtypes had slow Go RT. The ADHD-C group had a modest but significant deficit (F1,81 = 4.39, p = .039, η2 = 0.052). The ADHD-I group had a slightly larger response speed deficit (F1,53 = 5.46, p = .023, η2 = 0.093). However, the two ADHD groups did not differ significantly (p = .18, η2 = 0.03; Table 2). Planning: Tower of London Total Points
The group ⫻ gender interaction was nonsignificant (p = .91). When boys and girls were combined, the effect of diagnostic group was significant (F2,99 = 6.49, p = .002, η2 = 0.12). Compared with the control group, the ADHD-C group had a large deficit on the Tower of London (F1,82 = 15.7, p < .001, η2 = 0.16). The ADHD-I group had a
TABLE 2 Means (SDs) on Neuropsychological Measures for Three Groups Group Dependent Measure Stop Task Stop Signal RT Stop Task Go RT Tower of London points Stroop Word Stroop Color Stroop Color-Word Trailmaking A time Trailmaking B time
Control
ADHD-C
ADHD-I
297.7 (82.8) 654.2 (129.1) 28.7 (2.7) 74.2 (12.1) 50.3 (8.9) 29.8 (9.3) 18.3 (6.9) 41.3 (19.0)
415.3 (140.7) 705.6 (104.4) 26.6 (2.8) 60.4 (12.1) 43.0 (8.7) 24.2 (7.8) 22.0 (10.3) 56.6 (36.8)
382.2 (200.1) 732.5 (107.2) 27.0 (3.1) 60.3 (13.2) 43.9 (11.0) 23.0 (5.9) 25.3 (12.1) 49.6 (21.4)
Note: ADHD-C = attention-deficit/hyperactivity disorder combined type; ADHD-I = ADHD inattentive type; RT = reaction time. For Stop Signal inhibition, means (SDs) for boys and girls were as follows. Controls: boys = 303.3 (89), girls = 292.0 (72); ADHD-C: boys = 426.4 (138), girls = 396.7 (126); ADHD-I: boys = 323.9 (177), girls = 471.3 (144).
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
63
NIGG ET AL.
smaller deficit that was shy of significance (F1,54 = 3.55, p = .08, η2 = 0.06); it was reduced further with IQ covaried (p = .13). The ADHD subtypes did not differ (p = .69, η2 < 0.01). Interference Control: Stroop Test
For the Color-Word condition, the group ⫻ gender interaction was nonsignificant (p = .24). Collapsing across gender, the main effect of diagnostic group was significant (F2,99 = 5.84, p = .004, η2 = 0.11). Simple effects revealed impairment with a modest deficit on Color-Word Naming for ADHD-C (F1,81 = 4.96, p = .029, η2 = 0.06) and for ADHD-I (F1,53 = 9.04, p = .004, η2 = 0.15), but the two subtypes did not differ significantly (p = .25, η2 = 0.02). Similar effects held for Word and Color Naming (Table 2). However, the most important question was whether we would find a specific interference effect, which we evaluated with a 2 ⫻ 2 repeated-measures ANCOVA of Color-Word and Color Naming speeds. All interaction terms were well short of significance (Fs < 1.0), with effect sizes in the trivial range (η2 < 0.01), indicating that neither ADHD subtype had a specific deficit in interference control apart from their shared deficit in slow naming speed. Results were the same for the interference score (Golden, 1978) and the residual Color-Word score regressed on Color Naming. Set-Shifting: Trailmaking
Group ⫻ gender interactions were nonsignificant for both Trailmaking A and Trailmaking B (both ps > .45). Collapsing across gender, Trailmaking B time failed to yield a significant main effect for diagnosis (p = .21, η2 = 0.03; Table 2). However, the simple output speed test, Trailmaking A, yielded a significant diagnostic group effect (F2,99 = 3.71, p = .028, η2 = 0.07). On that task, the ADHD-C group, as expected, showed no significant deficit (p = .33, η2 = 0.012), consistent with the prior literature (Pennington and Ozonoff, 1996). The ADHD-I group, in contrast, showed a sizeable deficit (F1,54 = 7.76, p = .007, η2 = 0.13) that remained significant with IQ covaried (p = .01) but fell shy of significance (p = .06) with RD covaried. The ADHD-C versus ADHD-I comparison was initially shy of significance (p = .06, η2 = 0.06) but was significant with IQ covaried (F1,58 = 4.31, p = .042, η2 = 0.07). DISCUSSION
This study investigated two ADHD subtypes on four operational domains of executive functioning and on out64
put speed. Overall, the findings are consistent with the literature, but they also introduce new considerations for clarifying the relation of the two ADHD subtypes. The two ADHD groups had similar deficits compared with controls on key response speed measures, such as speed of the Go response (Stop task) and naming speed on the Stroop subtests, suggesting shared deficits in effort or vigilance (Sergeant et al., 1999). Only the ADHD-C group had a deficit in planning, but even on that measure, the two ADHD subtypes did not differ significantly from one another. The finding that, on most comparisons, ADHD subtypes did not differ significantly was also reported by Houghton et al. (1999), as well as by Hinshaw et al. (under review) in a study of girls. The findings suggest that these two subtypes are related by virtue of some shared neuropsychological deficits. However, this conclusion depends on consideration of executive function component functions and gender. First, the two ADHD subtypes did differ on the Stop task. Our findings break new ground in comparing the two subtypes on motor inhibition. This Stop task is of substantial theoretical importance because of its isolation of a motor inhibition process, which is crucial to theories of ADHD (Barkley, 1997) but is not well-assessed in most executive tasks. Effects, however, depended on gender. For the ADHD-C group, both boys and girls exhibited deficits compared with controls. In addition, ADHD-C boys were impaired compared with ADHDI boys, supporting a mechanism-based distinction between the subtypes for boys with regard to motor inhibition. For the ADHD-I group, boys performed nearly normally on the Stop task, whereas girls were impaired (versus controls; see means in Table 2 footnote). Second, the ADHD subtypes differed on Trailmaking A, in which the ADHD-I group was impaired compared with the ADHD-C group, with IQ covaried. This finding is notable mainly because it supports the conclusion from the other measures that, like ADHD-C, ADHD-I is also characterized by slow motor output speed or impaired vigilance processes. One study failed to find a deficit in ADHDI on Trailmaking A (Houghton et al., 1999); the effect may depend on group selection procedures. Our failure to find deficits in ADHD-C on Trailmaking B echoes the prior literature (Pennington and Ozonoff, 1996). The finding of an ADHD-C (but not ADHD-I) deficit on planning replicates the findings of Klorman et al. (1999), although the direct subtype comparison was not significant in our smaller sample. We did not find a deficit in J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
ADHD SUBTYPES AND EXECUTIVE FUNCTIONS
either ADHD group in the interference function of the Stroop task apart from naming speed. Many Stroop studies of ADHD failed to report the task ⫻ group interaction or interference score (e.g., Houghton et al., 1999). Like us, those that did so failed to find a significant ADHD deficit in children (e.g., Grodzinsky and Diamond, 1992). However, an interference control deficit has been reported in adolescents (MacLeod and Prior, 1996). Although our results were unchanged with reading ability covaried, more examination of the possible role of subclinical reading difficulties in concealing an ADHD interference control deficit in childhood will be of interest. By adolescence, even slow readers may show the interference effect in ADHD.
appeared to be a parallel between rated symptoms of inattention and neuropsychological deficits in output speed. Conclusion
Results inform neuropsychological executive theories of ADHD subtype dysfunction. The data suggest that ADHD-C and ADHD-I are related disorders that share deficits in vigilance or effort functions and, on many measures, may differ only in severity. However, at least in boys, they may be distinguishable on the basis of response inhibition deficits found only in the ADHD-C group. REFERENCES
Limitations
Sample size was not equal for the two ADHD subtypes, so statistical tests may underestimate the deficits in the ADHD-I versus the ADHD-C group, or the small ADHD-I group could yield chance findings. In addition, because we did not intend to focus on gender differences, the number of girls in the study was small; clarification of gender effects will require a larger sample of boys and girls. Although the “or” algorithm maintains fidelity to the validity data in the DSM-IV field trials, it lowers the threshold for identifying both combined and inattentive subtypes versus other (albeit less well validated) methods of totaling the symptoms. Children on long-acting medications were allowed into the sample (with no change in results); this may be a difference from some other ADHD samples. Also, it should be noted that the executive constructs were generally assessed with only one measure each; convergence across measures will be important to examine in future work. Clinical Implications
The sample was recruited in the community. Although inferential advantages support such sampling, and most ADHD children had received clinical services, confirmation that findings extend to clinic-recruited samples is needed. From a clinical perspective, the possibility that children with ADHD-I have deficits in response speed (indexing effort or arousal effects) that are similar to those of children with ADHD-C underscores the importance of recognizing that children with ADHD-I might have notable cognitive deficits that should be assessed. Whereas the ADHD-I group appears as a milder behavioral variant compared with the ADHD-C group, as a result of less severe disruptive symptoms, in both groups there J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2
Achenbach TM (1991), Manual for the Child Behavior Checklist/418 and 1991 Profile. Burlington: University of Vermont Department of Psychiatry Barkley RA (1997), Behavioral inhibition, sustained attention, and executive function: constructing a unified theory of ADHD. Psychol Bull 121:65–94 Barkley RA, Grodzinsky GM (1992), Are tests of frontal lobe functions useful in the diagnosis of attention deficit disorders? Clin Neuropsychol 8:121–139 Cabeza R, Nyberg L (1997), Imaging cognition: an empirical review of PET studies with normal subjects. J Cogn Neurosci 9:1–26 Cantwell DP, Baker L (1992), Attention deficit disorder with and without hyperactivity: a review and comparison of matched groups. J Am Acad Child Adolesc Psychiatry 31:432–438 Carlson CL, Mann M (2000), Attention deficit hyperactivity disorder, predominantly inattentive subtype. Child Adolesc Psychiatr Clin N Am 9:499–510 Castellanos FX (1999), The psychobiology of attention-deficit/hyperactivity disorder. In: Handbook of Disruptive Behavior Disorders, Quay HC, Hogan AE, eds. New York: Kluwer/ Plenum, pp 179–198 Cohen J, Cohen P (1983), Applied Multivariate Regression/Correlation Analysis for the Behavioral Sciences, 2nd ed. Hillsdale, NJ: Erlbaum Conners CK (1997), Conners Rating Scales-Revised. Toronto: Multi-Health Systems Doyle AE, Biederman J, Seidman LJ, Weber W, Faraone SV (2000), Diagnostic efficiency of neuropsychological test scores for discriminating boys with and without attention deficit-hyperactivity disorder. J Consult Clin Psychol 68:477–488 Faraone SV, Biederman J, Weber W, Russell RL (1998), Psychiatric, neuropsychological and psychosocial features of DSM-IV subtypes of attention-deficit/hyperactivity disorder: results from a clinically referred sample. J Am Acad Child Adolesc Psychiatry 37:185–193 Golden CJ (1978), Stroop Color and Word Test: A Manual for Clinical and Experimental Uses. Chicago: Stoelting Grodzinsky GM, Diamond R (1992), Frontal lobe functioning in boys with attention deficit hyperactivity disorder. Dev Neuropsychol 8:427–445 Houghton S, Douglas G, West J et al. (1999), Differential patterns of executive function in children with attention-deficit hyperactivity disorder according to gender and subtype. J Child Neurol 14:801–805 Keppel G (1991), Design and Analysis, 3rd ed. Englewood Cliffs, NJ: Prentice Hall Klorman R, Hazel-Fernandez LA, Shaywitz SE et al. (1999), Executive functioning deficits in attention-deficit hyperactivity disorder are independent of oppositional defiant disorder or reading disorder. J Am Acad Child Adolesc Psychiatry 38:1148–1155 Krikorian R, Bartok J, Gay N (1994), Tower of London procedure: a standard method and developmental data. J Clin Exp Neuropsychol 16:840–850 Lahey BB, Applegate B, McBurnett K et al. (1994), DSM-IV field trials for attention-deficit/hyperactivity disorder in children and adolescents. Am J Psychiatry 151:1637–1685
65
NIGG ET AL.
Levin HS, Mendelsohn D, Lilly MA et al. (1994), Tower of London performance in relation to magnetic resonance imaging following closed head injury in children. Neuropsychology 8:171–179 Logan GD, Schachar RJ, Tannock R (1997), Impulsivity and inhibitory control. Psychol Sci 8:60–64 Lyon GR, Krasnegor NA (1996), Attention, Memory and Executive Function. Baltimore: Paul H. Brooks Publishing Company MacLeod CM (1991), Half a century of research on the Stroop effect: an integrative review. Psychol Bull 109:163–203 MacLeod D, Prior M (1996), Attention deficits in adolescents with ADHD and other clinical groups. Child Neuropsychol 2:1–10 Nigg JT (1999), The ADHD response inhibition deficit as measured by the Stop Task: replication with DSM-IV combined type, extension, and qualification. J Abnorm Child Psychol 27:391–400 Nigg JT (2001), Is ADHD an inhibitory disorder? Psychol Bull 127:571–598 Pennington BF (1997), Dimensions of executive functions in normal and abnormal development. In: Development of the Prefrontal Cortex: Evolution, Neurobiology, and Behavior, Krasnegor NA, Lyon GR, Goldman-Rakic P, eds. Baltimore: Paul H Brooks Publishing Company, pp 265–281 Pennington BF, Ozonoff S (1996), Executive functions and developmental psychopathology. J Child Psychol Psychiatry 37:51–87 Posner MI, Peterson SE (1990), The attention system of the human brain. Annu Rev Neurosci 13:25–42 Quay HC (1997), Inhibition and attention deficit hyperactivity disorder. J Abnorm Child Psychol 25:7–13 Reitan RM (1979), Manual for Administration of Neuropsychological Test Batteries for Adults and Children. Tucson, AZ: Author Reynolds CR, Kamphaus RW (1992), Behavioral Assessment System for Children Manual. Circle Pines, MN: American Guidance Service
66
Rovine MJ, Delaney M (1990), Missing data estimation in developmental research. In Statistical Methods in Longitudinal Research, Vol 1, von Eye A, ed. San Diego: Academic Press, pp 35–79 Sattler JM (1992), Assessment of Children: Revised and Updated 3rd Edition. San Diego: Author Seidman LJ, Biederman J, Faraone SV, Weber W (1997), A pilot study of neuropsychological function in girls with ADHD. J Am Acad Child Adolesc Psychiatry 36:366–373 Sergeant JA, Oosterlaan J, van der Meere J (1999), Information processing and energetic factors in attention deficit/hyperactivity disorder. In: Handbook of Disruptive Behavior Disorders, Quay HC, Hogan A, eds. New York: Kluwer/Plenum, pp 75–104 Shaffer D, Fisher P, Lucas C (1997), NIMH Diagnostic Interview Schedule for Children-IV. New York: Ruane Center for Early Diagnosis, Division of Child Psychiatry, Columbia University Swanson JM, Lerner MA, March J, Gresham FM (1998), Assessment and intervention for attention-deficit/hyperactivity disorder in the schools: lessons from the MTA study. Pediatr Clin North Am 46:993–1009 Wechsler D (1991), Wechsler Intelligence Scale for Children, 3rd ed. San Antonio, TX: Psychological Corporation Wechsler D (1992), Wechsler Individual Achievement Test. San Antonio, TX: Psychological Corporation Wilcox RR, Keselman HJ, Kowalchuk RK (1998), Can tests for treatment group equality be improved? The bootstrap and trimmed means conjecture. Br J Math Stat Psychol 51:123–134 Zametkin AJ, Rapoport JL (1987), Neurobiology of attention-deficit disorder with hyper-activity: where have we come in 50 years? J Am Acad Child Adolesc Psychiatry 26:676–686
J . A M . A C A D . C H I L D A D O L E S C . P S YC H I AT RY, 4 1 : 1 , J A N U A RY 2 0 0 2