Neurocognitive markers in late-onset ADHD have not been longitudinally investigated. This 6-year follow-up study aimed to determine whether children with late-onset ADHD could be differentiated from their siblings without ADHD by their neurocognitive functioning at baseline and at follow-up.

The 6-year longitudinal study utilised a sub-sample of the Dutch International Multicentre ADHD Genetics (IMAGE) cohort. Several neurocognitive functions (timing, reaction time [variability], motor control, working memory and intelligence) were measured at baseline and at the follow-up. There were four different groups involved in the study: probands who had a childhood-onset (persistent) ADHD diagnosis based on Diagnostic and Statistical Manual (DSM-IV) criteria of combined-type ADHD at baseline and meeting full DSM-5™ criteria of ADHD at follow-up; siblings with late-onset ADHD who were unaffected in childhood but met full DSM-5™ criteria at follow-up; siblings without ADHD who did not meet criteria for ADHD at either baseline or follow-up; and controls who were individuals that did not meet criteria for ADHD at either baseline or follow-up.* Neurocognitive variables were identically measured at baseline and follow-up. IQ, gender and follow-up time were used as covariates.

There were 524 participants (childhood-onset ADHD, n=193; sibling with late-onset ADHD, n=34; siblings without ADHD, n=111; controls, n=186) included in this study out of the original IMAGE sample of 1069 participants (von Rein et al, 2015). Age at baseline was approximately 11 years, age at follow-up was approximately 17 years. At baseline, siblings with late-onset ADHD did not differ significantly in their IQ (mean [standard deviation; SD]: 104 [11]) compared with siblings without ADHD (105 [11]) and controls (107 [11]). Healthy controls had a higher IQ compared with childhood-onset ADHD at baseline (p < 0.001) and follow-up (p < 0.001). Furthermore, the interaction between age, neurocognitive function and group was significant for childhood-onset ADHD versus siblings without ADHD (b = –0.03, p = 0.004) for motor timing variability. Analyses in two equal-sized age groups (aged <12 years and ≥12 years) revealed that the motor timing variability (the interaction effect between group and time) was only significant in the youngest age group (p < 0.006).

Baseline neurocognitive functioning and symptoms

Based on the Conners’ Parent Rating Scale-Revised (CPRS-R), parents reported that total symptom severity was significantly higher in siblings with late-onset ADHD compared with siblings without ADHD (mean [SD]: 57.1 [15.6] vs 47.8 [5.7]; p < 0.001). Furthermore, siblings with late-onset ADHD had significantly higher ADHD scores than the controls but ADHD scores of siblings with late-onset ADHD at baseline did not score within the clinical range (total ADHD [T] ≥ 63). Siblings with late-onset ADHD had significantly lower total ADHD symptom severity compared with individuals with childhood-onset ADHD at baseline. Neurocognitive functioning was measured through reaction time speed and motor control precision. There were no significant group differences for reaction time speed (p = 0.15) or motor control precision (pursuit task; p = 0.44) at baseline. Siblings without ADHD differed significantly on measures of time reproduction precision (p < 0.001), reaction time variability (p = 0.001), time production variability (p < 0.001), motor control precision (p ≤ 0.001) and verbal working memory (p < 0.001) at baseline. However, pairwise comparison showed that there was no significant difference between siblings with late-onset and childhood-onset ADHD on any of the neurocognitive measures and their performance was worse compared with the controls at baseline. On the other hand, the control group and siblings without ADHD group equally performed better than the late-onset and childhood-onset ADHD groups on reaction time variability and time production variability. The same pattern was observed in a motor control precision task (tracking task) but there was no significant difference between siblings with late-onset ADHD and siblings without ADHD. No significant differences were observed between siblings with late-onset ADHD, childhood-onset ADHD and siblings without ADHD after pairwise comparisons of verbal working memory and time reproduction precision. However, these three groups all performed worse than the control group.

Follow-up neurocognitive functioning and symptoms

No significant difference in total symptom severity was reported by parents or teachers at follow-up between siblings with late-onset ADHD and childhood-onset ADHD. However, they both had significantly higher total symptom severity compared with siblings without ADHD and healthy controls. Emotional lability was reported by parents to be significantly higher in siblings with late-onset ADHD (mean [SD]: 52.9 [9.1]) compared with siblings without ADHD (45.7 [5.5]), but there was no significant differences in anxiety between the two groups (51.3 [9.6] vs 49 [8.0]). Additionally, there were similar levels of emotional lability between siblings with late-onset ADHD (52.9 [9.1]) and childhood-onset ADHD (57.1 [14.0]) but were not in the clinical range (T ≥ 63) at follow-up. There were significant differences between groups for all of the neurocognitive functioning tasks: time reproduction precision (p = 0.001); reaction time speed (p = 0.01); reaction time variability (p = 0.002); motor control precision (pursuit task, p < 0.001; tracking task, p = 0.001); verbal working memory (p < 0.001); and time production variability (p < 0.001). There were no significant differences between siblings with late-onset and childhood-onset ADHD in the neurocognitive measures after pairwise comparison at follow-up. However, both groups performed worse compared with the control groups on all neurocognitive measures. On the other hand, siblings without ADHD performed similarly to the control groups on all neurocognitive measures, except for motor control (tracking task). Motor control (tracking task) pairwise comparison showed there was no significant differences between siblings with late-onset, childhood-onset ADHD and stable unaffected ADHD, and they all performed worse than the control group. Additionally, siblings with late-onset ADHD performed worse on the time production variability and time reproduction precision measures compared with siblings without ADHD.

Neurocognitive change over time and sensitivity analyses

The authors determined there was no significant group by time interaction for all neurocognitive measures: time reproduction precision (p = 0.08), reaction time variability (p = 0.88), time production variability (p = 0.56), reaction time speed (p = 0.38), motor control precision (pursuit task, p = 0.15; tracking task, p = 0.15) or verbal working memory (p = 0.10). Additionally, when individuals were excluded who had anxiety (n=4) and/or emotional lability (n=5) scores above the clinical range (T ≥ 63) and/or with substance-use disorders (n=8) and/or nicotine dependency (n=7), findings were replicated. The authors stated that similar or comparable (non)significance levels and effect sizes were obtained.

There were several limitations to this study. Firstly, the sample size of the siblings with late-onset ADHD was relatively small, which may have impacted statistical power to detect small group differences. Secondly, the authors were unable to include all neurocognitive domains that are currently regarded as important in ADHD (e.g. reward-related neurocognitive functions). Finally, as the follow-up assessment occurred in young adulthood, there is the possibility that new cases of late-onset ADHD may have developed later.

To conclude, the study showed that siblings with late-onset ADHD had comparable neurocognitive deficits as individuals with childhood-onset ADHD. The authors suggested that the data may support the view that protective factors in childhood, such as higher intelligence, may mask or compensate for ADHD symptoms and impairments. Additionally, the data suggests that individuals with late-onset ADHD may share neurodevelopmental aetiology with childhood-onset ADHD. The authors stated that, due to the findings, siblings with late-onset ADHD have a milder clinical picture and broader externalising symptoms than individuals with childhood-onset ADHD, which puts them at risk of being overseen and under-recognised. Therefore, clinicians should be aware that individuals with late-onset ADHD have a less typical ADHD course after age 12 years than individuals with childhood-onset ADHD.

Read more about neurocognitive markers of late-onset ADHD here

*Inclusion criteria for entry at baseline were Caucasian descent, IQ ≥ 70, no diagnosis of autism, epilepsy, general learning difficulties and known genetic disorders