Attention Deficit/Hyperactivity Disorder (ADHD) is a neurodevelopmental disorder with a typical onset in preschool years. This period is important for cognitive development. For instance, elementary forms of Executive Functions (EF) components, such as working memory (WM) and inhibition, are already emerging before the age of three and develop rapidly in preschool years [1]. Language also develops at a rapid pace in this period and is often characterized by a transition from two word utterances in toddlerhood to complex sentences in late preschool years [2]. Children with high levels of ADHD symptoms often show delays or impairments in cognitive functions such as WM, inhibition, and language [3–9]. These delays are often present from an early age and can have a substantial impact on children’s academic and social functioning [10–12]. However, our understanding of associations between ADHD symptoms and cognitive functions in preschoolers is limited and mainly based on parent reports. Teacher reports of language skills would be particularly useful as the informants usually have a large “reference base” for comparison, from years of experience with children of the same age, representing the full range of language skills. Teacher reports provide information about language skills that may be challenging to elicit in a test situation with young preschoolers due to shyness and/or limited motivation or attention [13] and they are easy and inexpensive to administer.

Children with ADHD have been shown to have lower IQ-scores [14]. The term “IQ-score” refers to the performance on a limited set of standardized, cognitive tests which were originally combined to measure mental age, whereas “cognitive functions” is a broader term and refers to several specific cognitive functions, such as memory and language skills. Intellectual ability contributes to mental health already in young childhood. Four year old children with below average IQ-scores have been found more likely to experience mental health problems at school age [15]. Recent studies further indicate that associations between ADHD and cognitive functions depend on IQ and state that IQ-score therefore should be taken into account when investigating cognitive functions in ADHD [16, 17].

A number of studies have reported inconsistent findings regarding deficits in WM and inhibition in preschoolers with ADHD [18]. For example, weaker associations between cognitive functions and ADHD symptoms have been found in preschoolers than those typically found in older children [19] or no interaction between ADHD symptoms and IQ-score on either verbal or nonverbal WM have been reported [20].

In older children, a strengthening of associations between inattention, verbal WM, and verbal memory has been shown to occur through the school years [21, 22]. Contrary to this, it has also been suggested that children with ADHD have a developmental delay of verbal WM only until eight years of age [23]. It is unclear why these results are showing a contrary age-effect regarding verbal WM compared to previous studies, however, differences in cognitive measures and in sample characteristics might play a role.

Combined, findings from most preschool- and school aged samples raise the question if associations between ADHD symptoms and WM are strengthened with age. Cognitive functions must have reached a certain level of maturity before delays or deficits are visible and executive aspects of WM are just emerging at the age of three [24, 25]. Based on previous research and a theoretical assumption, certain hypotheses about the influence of IQ-score on the association between ADHD symptoms and cognitive functions in preschoolers can be made. Since stronger associations between ADHD symptoms and verbal memory have been found with increasing age in several studies and since young children with higher IQ-scores may have equivalent cognitive functioning as slightly older children with average intellectual development, it is possible that an association between ADHD symptoms and verbal WM grows stronger with increasing mental age/IQ-score. Hence, IQ-score could moderate the association between ADHD symptoms and nonverbal WM. Put differently, an interaction between IQ-score and ADHD symptoms on verbal WM might be expected: the higher the IQ-score, the stronger the association between ADHD symptoms and verbal WM. On the other hand, according to Sowerby and colleagues [23], weaker associations between ADHD symptoms and verbal WM with increasing IQ-score could be expected. There is also another reason for expecting a decrease in associations. As verbal WM is closely associated with language [26, 27], which in turn is closely associated with IQ-score as verbal ability is an important aspect of intellectual ability, verbal WM may be well-developed in children with high IQ-scores regardless of ADHD symptoms. In fact, weaker associations between ADHD symptoms and verbal WM with increasing IQ-score can also be expected with regard to nonverbal WM, as nonverbal WM is assumed to be language based as well [28]. However, not all studies have found support for age-related associations between ADHD and nonverbal WM [29]. In the light of these conflicting findings, in which methodological differences may explain some of the inconsistency, we leave predictions about a possible interaction between IQ and ADHD symptoms on WM open.

An interaction between ADHD symptoms and IQ-score on response inhibition is not likely, as impaired inhibition is closely related to ADHD symptoms [4, 5], and considerably less so with IQ-score [30]. ADHD symptoms may therefore be associated with impaired response inhibition regardless of IQ-score. On the other hand, inhibition has been found to be more strongly associated with ADHD symptoms in young school-aged children than in older, which would imply decreasing associations between ADHD symptoms and inhibition with age [22], however, we do not know whether this process is linear and whether it already starts in preschool years.

We would expect weaker associations between ADHD symptoms and preschool teacher-reported language skills with higher IQ-score, for the same reason as expecting weaker associations between ADHD symptoms and verbal WM with increasing IQ-score. As teacher-reported expressive language skills are closely related to intellectual ability [31], high IQ-score may be associated with well-developed expressive language skills regardless of ADHD symptoms in the present study, while children with lower IQ-scores may struggle with language, making an interaction between ADHD symptoms and IQ-score on preschool teacher-reported expressive language skills likely.

In sum, a higher IQ-score may either make cognitive problems related to ADHD symptoms more visible or it may act as a protector from the impact of such cognitive deficits, or there may be no effect of a high IQ-score on the association between ADHD symptoms and cognitive functions in preschoolers. In order to investigate these possible interactions, nonclinical samples should be used, minimizing the risk of studying samples with disproportionally few children with well-developed cognitive functioning and/or higher IQ-scores. The use of nonclinical samples may be particularly important when studying preschoolers, as the classification of ADHD in young children is challenging. Yet to our knowledge, there are no studies with nonclinical samples of preschoolers to date specifically investigating the impact of IQ-score on the association between ADHD symptoms and cognitive functions.

The main aim of this study was to test whether IQ-score influences associations between ADHD symptoms and WM, response inhibition, and reported expressive language, respectively, using a nonclinical sample of preschoolers. In order to address the main aim, we first investigated if there were significant associations between ADHD symptoms and the selected cognitive functions and in what direction they went and then went on to test whether there were significant interactions between ADHD symptoms and IQ on the cognitive functions mentioned above. A significant interaction would mean that the association between ADHD symptoms and the cognitive function under study depended on the level of IQ-score. To increase the clinical value of the results, follow-up analyses investigated whether negative associations between ADHD symptoms and selected cognitive functions differed in children with below median IQ-score compared with children with above median IQ-score.

The present study is a sub study of a longitudinal study of ADHD in preschool children. The longitudinal study recruited its participants from the Norwegian Mother and Child Cohort Study (MoBa) [32], a population based birth cohort at the Norwegian Institute of Public Health. The parents of the participating children gave informed consent to participate in the research and to the publication of the results. The present study was approved by the Norwegian Regional Ethics Committee for Medical and Health related Research.

Participants

The present study used a nonclinical sample of young preschool children recruited from the MoBa. A flowchart of the recruitment of the sample is presented in Figure 1.

In the MoBa, all parents received a questionnaire when their child was 36 months old. The questionnaire included 11 questions regarding hyperactivity, impulsivity and attention problems from the Child Behavior Checklist (six questions) [33] and diagnostic criteria for ADHD (five questions) [34], that were used to select participants. In order to oversample children with relevant symptoms, about 80% of invited children were those who scored at or above the 90^th percentile on these questions. Children were also invited to participate when hyperactivity was reported as a health problem. Children with severe medical conditions compromising their ability to conduct the clinical assessment and/or high levels of autistic symptoms were excluded from the study before invitations were sent (n = 149). A total of 2798 children were invited from the MoBa cohort during the period from August 2007 to January 2011on the basis of these criteria. Of these, 1048 children (37.5%) participated in the clinical assessments. An additional 654 randomly drawn children from the MoBa cohort were invited, of whom 147 (22.5%) participated. The total sample of children who were clinically assessed at the mean age of 42 months was therefore 1195. Of these, the present study included the 1181 children who had valid IQ-scores (618 boys and 563 girls), all aged between 37 to 46 months. Additional information was provided by parents and preschool teachers (see below).

Demographic characteristics and descriptive data are presented in Table 1.

The sample was very homogeneous regarding age and the children were mostly from families with highly educated parents. We estimated the annual income of the MoBa, from which the sample was drawn, to equal the mean income in Norway. The sample of the present study had more ADHD symptoms and ODD symptoms than one would expect to find in the normal population. The means and standard deviations of the cognitive measures of the whole sample were approximately as one would expect to find in the normal population.

Table 2 shows associations between ADHD symptoms, cognitive measures, and teacher-reported expressive language skills, controlled for IQ-score and gender.

Both IQ-score and ADHD symptoms were significantly associated with WM, response inhibition, and reported language skills. Girls outperformed boys on the measures of NVWM, response inhibition and on reported expressive language, while there were no gender differences on the VWM measure.

Table 3 shows interaction effects between ADHD symptoms and IQ-score on cognitive measures and teacher-reported expressive language skills.

There was a significant interaction effect between ADHD symptoms and IQ-score on teacher-reported expressive language skills, not on WM or response inhibition.

Figure 2 is a visual presentation of the significant interaction effect in Table 3. It illustrates how the association between ADHD symptoms and teacher-reported expressive language skills differed depending on intellectual ability level. According to Cohen, Cohen, West, and Aiken [43], plotting an interaction is the first step to its interpretation.

Furthermore, follow-up analyses were performed to test if the findings of no interaction between ADHD symptoms and IQ-score on either VWM, NVWM, or response inhibition were due to limited statistical power. These regression analyses controlled for gender showed that associations between ADHD symptoms and verbal WM and nonverbal WM, respectively, were of similar sizes in both IQ-divided groups (verbal WM: below median IQ group (n = 525): B = -0.07, SE B = 0.03, β = -.09, t = -2.09, p = .037, R² = .1; median or above median IQ group (n = 600): B = -0.10, SE B = 0.03, β = -.14, t = -3.48, p = .001, R² = .2; nonverbal WM: below median IQ group (n = 558): B = -0.13, SE B = 0.03, β = -.18, t = -4.43, p < .001, R² = .5; median or above median IQ group (n = 616): B = -0.09, SE B = 0.03, β = -.13, t = -3.15, p = .002, R² = .2). Associations between ADHD symptoms and response inhibition were also of similar sizes in both groups: below median IQ group (n = 482): B = -0.36, SE B = 0.10, β = -.17, t = -3.80, p < .001, R² = .07; median or above median IQ group (n = 549): B = -0.37, SE B = 0.10, β = -.16, t = -3.85, p < .001, R² = .05. However, associations between ADHD symptoms and reported expressive language skills were not similar in size in the two groups (below median IQ group (n = 507): B = -0.42, SE B = 0.09, β = -.21, t = -4.95, p < .001, R² = .07; median or above median IQ group (n = 560): B = -0.14, SE B = 0.07, β = -.08, t = -2.00, p = .046 R² = .01).

IQ-score and ADHD symptoms were significantly associated with WM, response inhibition, and teacher-reported language skills in our sample of children. Girls outperformed boys on most measures. ADHD symptoms had a differential effect on teacher-reported expressive language depending on IQ-score: in children with below median IQ-score, a larger number of ADHD symptoms were more likely to be accompanied by poorer teacher scores of expressive language skills, while the level of ADHD symptoms exerted a smaller effect on reported language skills in children with above median IQ-score. As for the other cognitive measures, no such interactions were found, as associations between ADHD symptoms and WM and response inhibition, respectively, were significant and of similar sizes regardless of IQ-score. Thus, our hypotheses regarding teacher reports of expressive language and response inhibition were supported.

We found no interaction between IQ-score and ADHD symptoms on either verbal WM or nonverbal WM. This could mean that the association between ADHD symptoms and WM is strengthened with increasing IQ-score in some children, while the association is weakened with increasing IQ-score in others. This interpretation would partially support previous studies [21–23], if the assumption that chronological and mental age show similar patterns is correct. We did not find support for the assumption that WM is well-developed in children with high IQ-scores regardless of ADHD symptoms, due to its close association with language [26, 27]. Instead, the results indicate that young preschoolers with high levels of ADHD symptoms can struggle with reduced WM regardless of IQ-score. The same conclusion was reached for response inhibition, as it was associated with ADHD symptoms regardless of IQ-score. Inhibition is assumed to be one of the first executive functions to develop [44], making inhibitory deficits a central feature of ADHD already at a young age [45].

Some aspects of the design of the present study must be discussed as IQ-score is interrelated with ADHD symptoms and with other cognitive functions. However, previous studies have shown that unique aspects of cognitive functions contribute to IQ-score [46], and that impaired executive functions, e.g. response inhibition and WM, do not account for lower IQ-scores in children with ADHD [47, 48]. Expressive language skills, on the other hand, are likely to be more closely connected with IQ-score. This may be particularly true for teacher-reports of language [31]. The finding of an interaction between intellectual functioning and ADHD symptoms on teacher-reported expressive language in the present study supports this finding [31]. Teacher-reported expressive language was included in the present study as it was measured independently from the other constructs under study (ADHD symptoms were reported by parents and WM and response inhibition were assessed with neuropsychological tests). It is also a convenient measure of language and could thereby facilitate more frequent considerations of language skills in ADHD assessments. Teacher reports of expressive language skills may be particularly useful when assessing young preschoolers as many of these tend to be shy when required to speak in direct assessments, whereas receptive language (which usually includes pointing at pictures in response to a question) is more easily assessed in this age group.

IQ-score was found to influence the association between preschool teacher-reported expressive language skills and ADHD symptoms. In young preschool children with higher IQ-scores, ADHD symptoms were more weakly associated with delayed expressive language than in children with lower IQ-scores. IQ-score did not influence associations between ADHD symptoms and WM and response inhibition, respectively.

Findings of the present study show that language skills of children should be considered when an ADHD assessment is undertaken, especially when intellectual ability is average or low. This is not routinely done, even though language delay can have considerable negative consequences at school age [53]. Furthermore, our findings indicate that preschool teachers and clinicians need to be aware that children with higher intellectual ability can still suffer from significant ADHD symptoms and related cognitive difficulties. The clinical implications of the present study should be viewed in the context of previous studies which have shown that neuropsychological assessments and early interventions are important for better prognosis for children with ADHD [54] and that the enhancement of neuropsychological functioning in preschool years holds promise for improving prognosis for children with ADHD [55]. Delays or impairments in WM, response inhibition, and expressive language affect communication and social interaction, both of which are related to prognosis [56].

Future studies are encouraged to consider the intellectual ability of their sample before generalizing findings about ADHD symptoms and teacher-reported language skills. The results of the present study suggest that associations between ADHD symptoms and teacher-reported language skills are more likely to occur in samples with lower IQ-scores than in samples with well-developed intellectual ability.

^aCorrections issued from the Publisher of the NEPSY test (Psykologiförlaget AB), 2002, were used.