Abstract
Attention deficit hyperactivity disorder (ADHD) is one of the most common psychiatric disorders of childhood. The prevalence in school age children is reported to be between 3% and 5% [1], and almost half of all child clinical referrals are for ADHD [2]. The specific aetiology of this disorder is not known, however, in recent years it has been argued that a dysfunction of the frontal lobes of the brain, in particular the fronto-striatal networks, underpins the symptoms of ADHD.
This hypothesis initially arose from the observed similarity in the presentation/behavioural symptoms of children diagnosed with ADHD and patients with frontal lobe damage [3, 4]. Lesions of the frontal lobe, particularly the prefrontal cortex are frequently associated with symptoms of hyperactivity, impulsivity, or inattention; the symptoms used to define ADHD [3–6]. In addition, structural [7–11] and functional [12–14] neuro-imaging studies have identified subtle abnormalities in the frontal networks of children with ADHD providing some direct evidence of frontal lobe involvement.
A number of measures thought to assess frontal lobe functioning have been tested on children with ADHD. These include tests of executive functions: selfregulation, sequencing of behaviour, flexibility of thinking, response inhibition, planning and organization of behaviour, attention, verbal and non-verbal fluency, and working memory; processes which the fronto-striatal networks are believed to control.
Studies using traditional neuropsychological tests of frontal/executive function [3, 6, 15–21], non-traditional executive function measures [22] and computerized test batteries [23–25] generally support the hypothesis that children with ADHD perform poorly on tests of frontal lobe functioning relative to healthy controls. However, not all studies offer strong support for frontal lobe deficits. Loge et al. [26] found that while children with ADHD performed more poorly than controls on tests of frontal lobe functioning their level of impairment was not as severe or global as reported in previous studies. Klorman et al. [27] and Houghton et al. [28] reported executive function deficits in children with combined type ADHD only, while Seidman et al. [29] failed to find evidence of frontal lobe deficits in a sample of girls with ADHD. The extent to which impaired frontal lobe performance persists into adolescence has also been questioned [8, 30, 31].
In summary the literature appears to support the hypothesis that children with ADHD suffer from frontal lobe dysfunction. However, the lack of consistent findings across studies makes it difficult to draw firm conclusions about the specific nature of the neuropsychological impairments associated with ADHD. It is likely that at least some of these inconsistencies are due to methodological differences between studies. Important among these differences is the criteria used to define ADHD, the medication status of participants and the attention given to the assessment of general intellectual functioning.
Published studies include children meeting criteria for DSM-III ADD/H [3, 19, 20, 32], DSM-III-R ADHD [3, 8, 15, 16, 18, 20, 26, 29, 33] and recently, DSM-IV ADHD [1, 22–24, 27, 28, 31] as well as those described as hyperactive [25] but not meeting DSM criteria for ADHD. Methods used to assess if children met these diagnostic criteria vary widely across studies. Another factor is stimulant medication, which is reported to improve executive function test performance [24]. A number of the studies reviewed fail to give consideration to this as a confounding variable. Some studies include mixed groups of children on and off medication [6, 15, 16, 29] while others do not report the medication status of their participants [8, 19, 25, 26].
Variation in the attention given to assessing general levels of intellectual functioning includes failing to assess general intellectual function [21], the use of inappropriate measures (e.g. Wechsler Adult Intelligence Scale [8]; Peabody Picture Vocabulary Test [3]), and the use of abbreviated versions of the Wechsler Scales [6, 15, 16, 20, 22, 25, 28] or other brief measures [31] which provide less stable IQ scores and poorer internal consistency [34]. As some studies have found that functions other than those subserved by the frontal lobes may be impaired in children with ADHD it is important to have an accurate assessment of general intellectual function to determine if the dysfunction is global, global but with relatively greater impairment of frontal functions, or restricted to functions under the control of the frontal lobes.
Mindful of these issues, the present study was based on a sample of children who met criteria for DSM-IV combined type ADHD, and who had never been treated with stimulant medication. The extent to which these children were impaired on tests of general intellectual functioning, as well as tests of frontal lobe/executive function, was investigated. This was achieved by comparing the performance of the children with ADHD on traditional neuropsychological measures of intelligence and frontal lobe function with an age and sex matched sample of children without behaviour problems.
Method
Participants
Data from 56 children, 28 diagnosed with DSM-IV combined type ADHD [1] and 28 normal controls, were included in the current study. There were 25 boys and 3 girls in each group ranging in age from 6 years 1 month to 10 years 11 months. The children in the control group were individually matched for age (± 3 months) and sex with the children in the ADHD group. Fifteen of the children in the ADHD group had a comorbid disruptive behaviour disorder (oppositional defiant disorder n = 9; conduct disorder n = 6).
Children in the ADHD group were referred to the study from the Department of Paediatrics (n = 16) and the Department Child, Adolescent and Family Mental Health (n = 10) at a local hospital and from the Learning and Guidance Unit of a local school (n = 2). Control participants were recruited by means of targeted newspaper advertisements and letters sent home to parents of children attending local schools.
All of the children included in the study had Wechsler Intelligence Scale for Children-third edition (WISC-III) Full Scale IQ scores of at least 70 and showed no evidence of neurological disorder or psychosis. At the time of testing none of the children in the ADHD group were prescribed medication for the management of ADHD.
Procedure
Ethical approval for the study was obtained from the Otago Ethics Committee. The parents of all children who participated gave written consent for their child to take part and the children gave verbal assent.
Prior to completing the neuropsychological assessment all of the children underwent a detailed diagnostic assessment that included parent, teacher, and child interviews together with parent and teacher completed behavioural questionnaires.
Diagnostic assessment
The parent interview obtained a detailed description of each child's behavioural difficulties together with his or her developmental and medical history. The children's teachers were interviewed, over the telephone, about each child's behavioural, academic, and social functioning. Each child was interviewed according to the Rutter and Graham Interview Schedule [35]. All interviews were conducted by three doctoral students in clinical psychology who were experienced in working with children and families. The students were trained to conduct these interviews by GT, a clinical psychologist experienced in assessing children with ADHD [36].
Parents and teachers completed the Disruptive Behaviour Disorders Rating Scale (DBDRS), which assesses for the presence and severity of DSM-IV symptoms of ADHD, oppositional defiant disorder and conduct disorder [37, 38]. Parents were asked to complete the Child Behavior Checklist (CBCL [39]), and the Conners 48-Item Parent Rating Scale 1 (CPRS [40]). Teachers were asked to complete the Teacher Report Form (TRF [41]), and the Conners 39-Item Teacher Rating Scale 1 (CTRS [40]).
Data from the interviews and the DBDRS were used in diagnosis. Children were diagnosed according to the DSM-IV criteria for ADHD. In applying these criteria the children were required to exhibit at least six symptoms of inattention and six symptoms of hyperactivity/impulsivity in one setting (home or school) together with evidence of symptoms in a second setting. The validity of the groups formed in this way was checked by comparing the T scores for the two groups on the Attention Problems scales from the CBCL and the TRF and the Hyperactivity Index T scores from the CPRS and CTRS (see Table 1). The two groups differed significantly on all four scales (Attention Problems: CBCL t(29.92) = 11.62, p < 0.001, TRF t(32.86) = 7.34, p < 0.001. Hyperactivity Index: CPRS t(41.74) = 14.39, p < 0.001, CTRS t(39.25) = 8.75, p < 0.001).
Demographic and behavioural characteristics of the Attention Deficit Hyperactivity Disorder (ADHD) and control groups
Neuropsychological assessment
Testing was carried out over two sessions on the same day in a quiet room in the ADHD Research Clinic at the University of Otago. Participants were given set breaks throughout testing to lessen the effects of fatigue. The neuropsychological tests were administered by graduate students in clinical psychology in the order indicated below.
Measures, administration, and scoring
The children completed the Australian adaptation of the WISC-III [42] and a battery of neuropsychological tests selected for their reported sensitivity to frontal lobe dysfunction. The following domains of functioning were assessed: verbal and non-verbal fluency, reasoning, problem solving, spatial working memory and attention.
The neuropsychological test battery included a Design Fluency task modelled on that of Jones-Gotman and Milner [43]. In the Free condition children were instructed to draw as many unnameable objects as they could in three minutes. Instructions for the Fixed condition were the same except that the children were told their drawings must be made up of four lines; the first two items from the Verbal Fluency Task from the British Abilities Scale [44]; booklet B of the Matrices Task from the British Abilities Scale [44]; the Wisconsin Card Sorting Test (WCST [45]); the Spatial Span test from the Wechsler Adult Intelligence Scale – Revised Neuropsychological Instrument [46]; the Auditory Continuous Performance Test (ACPT [47]); a Letter Cancellation Task modelled on that of Voeller and Heilman [19]. In this task the children were presented with three pages of letters in random order and given 10 min to cross off as many target letters (C's and E's) as they could; and the children's version of the Trail Making Test [48].
Data analysis
Prior to data analysis the distributions of the dependent variables were examined. Where extreme values were detected, the data highly skewed, or the assumption of homogeneity of variance violated the data were transformed. When data transformation did not correct extreme outlier values, but the assumptions for parametric analysis were met, the children with extreme scores were removed from the analysis. For analyses involving raw scores the data from the participants' agematched peers was also removed. Where there were numerous outliers or the data violated the assumptions of parametric analysis, non-parametric procedures were used. Proportion data were always arcsine transformed and other data were log transformed as required. Both the transformed and untransformed data are presented in the tables. Unless specified otherwise, all analyses were carried out using one way ANOVA.
Despite the number of comparisons made, for all analyses critical alpha was set at p < 0.05. While this increases the likelihood of type I errors, a more conservative alpha was not considered appropriate given the modest sample sizes and the potential for over correction when analysis of covariance was used (see Results). Effect sizes were calculated for all the neuropsychological test comparisons and are included in the tables.
Results
Demographic characteristics
The demographic and diagnostic characteristics of the ADHD and control group children are presented in Table 1. The two groups were well matched for age, sex, ethnicity and family structure. Mean family socioeconomic status was similar across the groups.
Wechsler Intelligence Scale for Children-III results
Summary scores for the two groups on the WISC-III are presented in Table 2. Significant group differences were identified for Full Scale IQ, F(1, 54) = 37.72, p < 0.001; Verbal IQ, F(1, 54) = 33.87, p < 0.001; and Performance IQ, F(1, 54) = 26.84, p < 0.001. Children in the control group obtained significantly higher IQ scores than children in the ADHD group.
Means, standard deviations, and effect sizes for the IQ, fluency, and reasoning tasks for the Attention Deficit Hyperactivity Disorder (ADHD) and control groups
Frontal lobe test results
The significant group difference in Full Scale IQ raises the question of whether IQ should be controlled for in analyses of the data from the tests of frontal lobe functioning. Opinions on this issue vary. Werry et al. [49] argue strongly in favour of controlling for IQ. On the other hand, Seidman et al. [16] suggest that controlling for IQ may remove variance directly attributable to ADHD. The correlation between IQ and ADHD severity is reported to be between r = −.30 and −.35 [2]. In the current study, frontal lobe test results were analysed with and without IQ as a covariate. Where significant group differences were detected and test performance correlated significantly with IQ, group comparisons were repeated with IQ as a covariate.
For tests for which age standardized scores were available, Full Scale IQ was correlated with test performance and used as the covariate. When raw scores were used in the analysis these were correlated with the sum of the raw scores on the WISC-III 2 (Raw IQ) and used as the covariate when necessary. This was done because correlating age corrected Full Scale IQ (FSIQ) scores with raw test scores has the potential to underestimate the relationship between IQ and frontal lobe test performance.
The means, standard deviations, sample sizes and effect sizes for the tests of frontal lobe functioning for the ADHD and control groups are presented in Table 2 (Design Fluency, Verbal Fluency, Matrices) and Table 3 (WCST, Spatial Span, Letter Cancellation, Trail Making and ACPT).
Means, standard deviations, and effect sizes for the problem solving, working spatial memory, and attention tasks for the Attention Deficit Hyperactivity Disorder (ADHD) and control groups
Design fluency task
In the Free condition the control group participants generated significantly more pictures than participants in the ADHD group F(1,54) = 5.32, p < 0.03. For the Fixed condition the control group also generated more pictures than the ADHD group, F(1,54) = 4.77, p < 0.04. Significant group differences were also found for the proportion of novel pictures and the proportion of errors, F(1,54) = 10.44, p < 0.003. Children in the control group produced a larger proportion of novel responses and a smaller proportion of errors. After covarying for raw IQ scores the proportion of novel and incorrect pictures approached significance F(1,53) = 3.87, p = 0.054.
Verbal fluency task
Responses to the two conditions were combined for the analyses. The proportion of errors were compared with the Mann–Whitney (MW) test. Significant group differences were found for the proportion of correct responses F(1,54) = 18.47, p < 0.001, the proportion of perseverative responses F(1,54) = 13.13, p < 0.02, and the proportion of errors MW = 258, df = 1, p < 0.002. Differences in the proportion of correct responses F(1,53) = 7.42, p < 0.01 and perseverative responses F(1,53) = 6.33, p < 0.02 remained significant after controlling for raw IQ scores.
Matrices task
On this task the control group produced significantly more correct designs than the ADHD group, F(1,54) = 5.62, p < 0.03. This difference was no longer significant after controlling for raw IQ scores.
Wisconsin Card Sorting Test
Significant group differences were identified for the error T-score F(1,41) = 4.80, p < 0.04; perseverative responses T-score, F(1,41) = 5.82, p < 0.03; perseverative errors T-score, F(1,41) = 5.06, p < 0.04; total number correct score, F(1,40) = 6.09, p < 0.02; and number of categories completed, F(1,40) = 5.39, p < 0.03. The control group performed better than the ADHD group on all of these measures. None of these differences remained significant after controlling for FSIQ for the T-scores and raw IQ scores for the number correct and number of categories.
Spatial span task
A 2 (group; ADHD vs control) × 2 (direction; forward vs backwards) repeated measures ANOVA was conducted with the Spatial span scores. This analysis indicated significant main effects for both direction and group. The children performed better on the forward than the backward section of the task F(1,50) = 9.28, p < 0.005, and the control group performed better than the ADHD group F(1,50) = 10.61, p < 0.003. The group by direction interaction was not significant. Following analysis of covariance, with raw IQ as the covariate, group differences were no longer significant.
Auditory Continuous Performance Test
Significant group differences were identified for the omission F(1,50) = 13.46, p < 0.001, and vigilance decrement scores F(1,50) = 6.90, p < 0.02. Children in the control group made fewer errors of omission and showed a smaller vigilance decrement than children in the ADHD group. These differences were no longer significant after controlling for raw IQ scores. χ2 analysis of the pass/fail data indicated that significantly more children with ADHD failed the task than would be expected by chance, χ2(1, 52) = 6.24, p < 0.02.
Letter cancellation task
The accuracy score, number of commission errors and lines completed were analysed with the non-parametric MW test. Children in the ADHD group made significantly more errors of commission than control group children, MW = 170, df = 1, p < 0.01. Raw IQ did not correlate significantly with scores on this task.
Trail making test
Raw scores from the Trail Making Test were converted to age standardized T scores using the norms presented in Spreen and Struass [50] Significant group differences were obtained for Trails A F(1,47) = 6.62, p < 0.02. The children in the control group completed the test more quickly. This difference was no longer significant after controlling for FSIQ.
Discussion
The present study compared the performance of children with combined type ADHD and an age and sex matched group of normal controls on measures of intellectual and frontal lobe functioning. Children with ADHD obtained significantly lower WISC-III IQ scores than the controls and did not perform as well on a battery of tests of frontal lobe functioning.
The poorer performance of the ADHD group on the WISC-III was not unexpected. Several studies have reported that children with ADHD obtain significantly lower scores than their age-matched peers on measures of intellectual functioning [15, 26, 51, 52]. However, contrary to previous reports the mean IQ of the children with ADHD in the current study was in the low average, rather than the average range [3, 8, 16, 27, 29].
There are a number of possible explanations for the lower scores obtained by the current sample of children with ADHD. The most obvious is the IQ cut-off score used in this study. Previous studies have typically excluded children with IQs below 80, whereas the current study excluded children with IQs below 70. This almost certainly lowered the mean IQ of the ADHD group. The poorer performance of the ADHD group may also reflect the nature of the sample. All of the children in the ADHD group were clinic referred and met criteria for DSM-IV combined type ADHD. These children may represent a more impaired sample than those included in earlier studies. Finally, all of the children in the ADHD group were medication naïve at the time of assessment. In addition to directly impacting on test performance, this may have influenced the children's prior learning opportunities, further contributing to their lower mean FSIQ.
Children in the ADHD group performed more poorly than controls across the range of tests of frontal lobe function. In general, the children in the ADHD group demonstrated less mental flexibility, poorer working memory, an increased error rate and increased perseverative responding compared to controls. These results are consistent with earlier investigations reporting that children with ADHD are impaired on tests sensitive to frontal lobe function [3, 6, 16–25].
Three alternative explanations for the poorer frontal lobe test performance of the ADHD group need to be considered. The test taking skills of the children with ADHD may be impaired by their behavioural or emotional problems. However, the testing situation contained those elements identified as likely to maximize the performance of children with ADHD (i.e. explicit rules, repetition of instructions, task novelty and high rates of reinforcement) [53]. In such situations ‘hyperactive children can be calm and attentive’ [22], p.404]. Given the significantly lower IQ scores obtained by the ADHD group, it is possible that differences on the frontal lobe tests reflect a generalized global impairment of cognitive functioning. This is unlikely because covariance analysis reduced, but did not completely eliminate, between group differences. After covariance analysis, the adjusted means continued to indicate poorer performance by the children in the ADHD group. Furthermore, covarying for IQ is likely to have removed some variance in test performance that is attributable to ADHD [16]. A more likely explanation for the poorer performance of the ADHD group is that these children do indeed suffer from impaired frontal lobe functioning. This explanation is certainly consistent with previous reports in the literature, however, it doesn't take account of the contribution to test taking performance of general intellectual functioning.
Together, the results from the WISC-III and the tests of frontal lobe function indicate that children with combined type ADHD have mild to moderate global cognitive impairment together with some impairment in functions subserved by the frontal lobes. The presence of impaired frontal functions raises the possibility that the generalized intellectual impairment may be a secondary effect caused by failure of the frontal lobes to mature normally. To test this hypothesis the cognitive development of children with ADHD needs to be carefully charted over time, evaluating general intellectual development alongside the development of functions of the frontal lobes.
The current study has a number of important strengths. First, the children in the control group were individually matched for age and sex to the target sample. This is important, as not all of the neuropsychological tests used are well normed for children. Second, none of the children in the ADHD group were taking, or had previously been prescribed, medication for ADHD. Third, all of the children in the study completed a rigorous diagnostic assessment using recommended procedures [54]. Finally, the inclusion of children with IQ scores of 70 and above increases the clinic representativeness of the sample of children with ADHD.
Notwithstanding these strengths a number of methodological issues need to be raised. The ADHD group was restricted to clinic referred children with combined type ADHD. This resulted in a well-defined sample with a similar symptom profile. However, it is not clear to what extent the current findings can be generalized to children presenting with inattentive or hyperactive/impulsive type ADHD, or those identified from community studies.
The ADHD group included children with comorbid oppositional defiant disorder and conduct disorder. While such comorbidity is consistent with the clinical presentation of ADHD, we cannot rule out the possibility that some of the differences between the ADHD and control groups are the result of these, or other, comorbid conditions. A number of recent studies suggest executive function deficits are independent of psychiatric comorbidity [6, 16, 27, 31, 55], but that comorbid learning disabilities may increase the severity of executive function deficits [6]. The inclusion of a psychiatric control group was beyond the scope of the current study and the services offered by the ADHD Research Clinic.
The neuropsychological tests were administered according to standard procedures, not computerized. As such, they are open to experimenter bias together with any language weaknesses among the children being tested, especially those with ADHD [25]. On the other hand, standard test administration allows for comparison with previous studies employing these test procedures.
Not all of the tests included in the frontal lobe battery were specifically developed for use with children, and are therefore not normed for this population. This limits test interpretation to some extent, and influenced the manner in which group differences in IQ were controlled.
Finally, numerous statistical tests were conducted, increasing the possibility of type I errors. However, effect sizes for the significant comparisons were all moderate to large and all comparisons were two-tailed despite research indicating children with ADHD are likely to perform less well on tests of frontal lobe function. With a larger sample it might be possible to decrease the number of comparisons by reducing the frontal lobe tests to their principal dimensions through factor analysis [31].
Clinical implications
The issue of whether the global impairment is a primary or a secondary problem in ADHD has important implications for the education of children with this disorder. If it is a primary problem then teaching assistance of a more global nature is likely to be necessary. If the generalized cognitive impairment is secondary to frontal impairment then early intervention to target the impaired frontal functions will be required to prevent the slippage in general intelligence. An obvious clinical recommendation arising from the current findings is the need for neuropsychological assessment and follow-up of children with ADHD to monitor their cognitive strengths and weaknesses and to assist in the planning and implementation of appropriate treatment and education plans.
Footnotes
Acknowledgements
The data collection was funded by grants from Lottery Health Research, the Health Research Council and the University of Otago, New Zealand. Thank you to the children who participated and members of the ADHD research team who assisted with data collection.
