A case-controlled, quantitative, experimental design was employed to establish whether children who screened positive for ADHD are impaired in WM and set-shifting. The sample was divided into participants classified as ADHD and matched controls without symptoms of ADHD.

Eleven primary schools from rural areas of the Limpopo province of South Africa were selected for the study.

Participants were Sepedi-speaking, Grade 1–7 learners, aged 6–15 years, whose teachers and parents completed the Disruptive Behaviour Disorders (DBD) rating scale^43,44 to identify children with ADHD symptoms. A total of 5451 learners were screened using the DBD rating scale. Children with scores above the 95th percentile were selected as the ADHD group.^15,42 Learners who scored below the 85th percentile formed the control group, in order to decrease the risk of false positives in the group. Children with scores of ≥ 26 on either or both the Inattentive and HI scales, matched for gender and age, were assigned to the ADHD group. These cut-off points were based on the results of the previous study by Meyer et al.,⁴¹ which was conducted on more than 6000 children in the Limpopo province and also employed the DBD. The control group, matched for gender and age, was established from the remaining cohort of children, who did not meet the inclusion criteria.

The resulting sample (N = 216) comprised 100 male and 116 female participants, and contained children who screened positive for ADHD symptoms (n = 108:50 males and 58 females) and a matched control group (n = 108:50 males and 58 females). Based on collateral information obtained from parents, teachers and school records, children with cerebral palsy, severe psychopathology, auditory and visual impairments, and intellectual disabilities were excluded from participation in the study.

This study was conducted in primary schools in the Lepelle-Nkumpi Municipality of the Capricorn district, Limpopo province, South Africa, during the year 2016. School principals provided consent for the study to be undertaken at their school. Informed, written consent was obtained from parents. The response rate for the DBD questionnaires was 96%. The testing procedure was explained to all learners selected for testing. The assessments were administered by clinical psychologists, with the help of field workers with psychology qualifications and who were conversant in Sepedi. Participants were individually assessed in a quiet, well-ventilated room. The test battery was administered during school hours, in accordance with the Department of Education’s regulations, in two sessions each lasting for approximately 60 min per participant. The total number of correct responses for the Digits subtest and the time taken to complete the Trail making test part B (TMT-B) were recorded.

Statistica 12.7 (Dell, 2015)⁵² was used for statistical analysis. MANOVA models were employed to examine probability between-group differences. Between-group differences were established using a 2 × 2 × 2 (ADHD group × gender × age) multifactorial ANOVA. Post hoc analysis (Bonferroni) was used to determine within-group differences (differences in gender and age).

Ethical clearance was obtained from the Biomedical Research Ethics Committee of University of KwaZulu-Natal, Durban, South Africa (reference number HSS/1446/015D) and the Department of Education, Limpopo province, South Africa. Principals of the selected schools provided their verbal consent for the study to be undertaken in their schools. Written, informed consent was obtained from parents or legal guardians of the learners to participate in the study. The researchers read out the assent form to children, and, after it was established that they had understood and agreed with the content, they were asked to assent to take part in the study. All children irrespective of age were treated in the same way. Voluntary participation was ensured, as participants were informed that they could withdraw at any stage should they want to do so. Parents of children who obtained positive scores were referred to the nearest hospital for further assessments and management.

Descriptive statistics, the MANOVA and the post hoc results for Digit Span are shown in Table 1. No main or interactions effect was found for gender. Statistical differences were found in performance between the ADHD and control group: Wilks L = 0.65, F = 56.15, p < 0.001, ƞp² = 0.35. A main effect for age was indicated: Wilks L = 0.90, F = 11.48, p < 0.001, ƞp² = 0.10, whilst no interactions effect was revealed for age. Post hoc analysis (Bonferroni) revealed that for both the Digits Forward and Digits Backwards, the older age group performed significantly better than the younger age group (both p < 0.001). An effect size (partial eta squared, ƞp²) of 0.35 (ADHD vs. control) can be considered as large and of 0.10 (difference in age) as medium.⁵³

Descriptive statistics, the ANOVA and the post hoc results for Trails-B are shown in Table 2. No main or interactions effect was indicated for gender. Statistical differences were found in performance between the ADHD and control group: F = 69.74, p < 0.001, ƞp² = 0.25. A main effect was indicated for age: F = 49.67, p < 0.001, ƞp² = 0.19, whilst no interactions effect for age was revealed. Post hoc analysis revealed that the older group performed significantly better than the younger age group (p < 0.001). Effect sizes (partial eta squared, ƞp²) of 0.25 (ADHD vs. control) and of 0.19 (difference in age) can be considered as large.⁵³

The results reveal that children with symptoms of ADHD, when compared with controls, showed significantly more WM deficits. This finding is in accordance with those of previous studies,^54,55 which suggested a significant difference between the two groups. The finding of this study supports Barkley’s EF model,¹⁵ which posits that children with putative ADHD experience difficulty retrieving information and keeping other details in mind whilst they attend to other activities. A study by Elosúa et al.³² also showed more WM deficits in children with possible ADHD, as indicated by lower scores on both the Digits Forward and Backward tests, when compared with the controls.

This study showed no gender differences in WM deficits in either the group with possible ADHD or the neurotypical controls. This finding contradicts previous findings by Skogli et al.⁵⁶ which showed higher rates of WM deficits in females with putative ADHD. The absence of gender interaction suggests that group differences in WM deficits may not be due to gender, as has previously been reported, where boys showed significant WM deficits when compared to girls.²⁷ This finding echoes some previous studies^26,28 that found no statistically significant gender differences in WM deficits amongst children with possible ADHD. A study by Piccardi et al.⁵⁷ also found no gender differences in WM amongst young adults (21–35 year olds) on DS, which suggests that in the case of persons who screened positive for ADHD, at least from childhood through to young adulthood, gender does not affect their ability to remember information over a short period of time, or to mentally utilise such information to learn, comprehend, and reason.

The findings of this study indicate the group independent age differences in WM functions, by showing that older children outperformed their younger peers on the Digits Span measurements, irrespective of whether they have possible symptoms of ADHD. This finding is contrary to some studies, which found that age did not affect the ability of children (viz. possible ADHD and controls) to sustain memory functions whilst engaging in cognitive activities.^26,27,34 The finding that younger children with or without putative ADHD displayed more deficits in WM suggests that the WM functions have possible neurodevelopmental basis associated with the maturation of the frontal lobe. Although it was not the primary focus of the current study, the fact that we could not find the group-related age differences in performance on functions of WM offers an explanation to previous research,^13,24 which reported that frontal lobe matures with age; however, the maturation does not depend on whether the person screened positive for ADHD symptoms. This finding is consistent with some previous studies,^23,58 which found that age significantly improved WM functioning in children with possible ADHD with regard to computerised tasks. In terms of age differences in WM functions amongst children with ADHD, the meta-analytical study by Kasper et al.²⁵ showed that WM functions tend to improve with age amongst children with possible ADHD and typically developing children aged 8–16 years, which suggests that although these functions are partly determined by environmental²² influences, the underlying developmental factors^7,59 and brain alterations that influence WM functioning²⁴ are similar across groups (viz. possible ADHD and Controls). This finding echoed that of Karalunas et al.,²³ which suggested that individual differences in visuospatial WM during the late childhood (7 through 13 years of age) predict ADHD symptom improvement as those who performed better in WM tasks also showed adequate improvement in ADHD symptoms. Consequently, older children with possible ADHD would be expected to show improved functioning, as they would have more time to developmentally ‘catch up’ to their typically developing peers.

Consistent with some previous studies,^32,34 this study found that children with possible ADHD performed poorly and made more errors on the TMT-B, and therefore, often required more time to complete the test when compared with the controls. This suggests that children with putative ADHD exhibit more deficits in set-shifting skills.^33,34 However, some studies found that children who screened positive for ADHD effectively sustain attention whilst switching between different domains.^28,30 Poor performance on the TMT indicates brain dysfunction in the prefrontal cortex, which is responsible for the set-shifting functions.³⁰

This study found no gender differences in set-shifting performance, which suggests that although behaviour is influenced by environment or culture, the neuropsychological mechanisms underlying behaviour are the same across genders.^8,24 This finding converges with that of other studies,^28,35 which reported that children who screened positive for ADHD, irrespective of gender, performed poorly on the TMT-B, and therefore, took more time to complete the task. Consistent with some previous studies,^28,26 older children performed better than the younger age group, who needed more time to complete the TMT-B. This finding departs from that of Rose et al.³⁷ who found no age differences in children’s ability to keep other matters in mind whilst attending to tasks.

Overall, these findings suggest that children with putative ADHD displayed more EF deficits, which supports Barkley’s model²¹ that children with possible ADHD have difficulty retrieving information from memory and keeping other details in mind whilst attending to different tasks.¹⁵ However, Minervino and Pereira⁶⁰ found non-significant group differences in WM and set-shifting, which suggests that individual deficits may not explain the multifaceted nature of ADHD. The contradictory results may imply that other psychological processes, which were not controlled in this study, may be involved. Whilst neuroimaging was not conducted, the results would be theoretically consistent with deficiencies in the dopamine reward valuation system.³⁸ In line with Seidman et al.,⁸ this study did not find gender differences in executive functioning, which suggests that whilst diverse environmental and/or cultural factors have impact on behaviour, possibly similar neuropsychological mechanisms underlie the executive functioning in children with putative ADHD across both genders. The fact that older children performed better than the younger age group could imply that ADHD and executive functioning are developmental processes.²⁸ In addition, the age of the sample may account for the lack of gender differences, as the gender-related differences may only be detected from the age of 12 years, with the onset of puberty,⁵⁹ when pubertal gonadal changes and related increased neural activation in the ventral striatum play a pivotal role in heightening ADHD symptoms.⁷ The ventral striatum is a part of the brain’s reward system that increases in typically developing pubertal samples, following positive reward, and is associated with heightened gonadal hormones.⁷

The implications of this study are that children with putative ADHD symptoms are significantly impaired in WM and set-shifting skills. The literature suggests that children with possible ADHD, and with deficits in WM²⁶ and set-shifting acuity,^33,35 often perform poorly at school when compared with the typically developing control group.^2,4 Earlier identification and intervention for children screening positive for ADHD and EF deficits can improve the functioning of these children.